Wikipedia talk:WikiProject Medicine/Archive 155

Archive 150

←

Archive 153

Archive 154

Archive 155

Archive 156

Archive 157

→

Archive 160

Automatic 'medical sources' links in Talk page headers is now live: your feedback is needed

Announcement: the {{Talk header}} template now incorporates the medical sourcing links from template {{find medical sources}} automatically, whenever a Talk page header is found on an article belonging to WP:WikiProject Medicine. The Talk header template generates the box you see at the top of some Talk pages.

When conditions are met, you will see the following section included in the box generated by the Talk header template, just above the archive links and search box (example shown is for "Premenstrual syndrome"):

This feature has now been released. On Talk pages belonging to WikiProject Medicine, you should now see source links like the ones illustrated above if the page contains a Talk page header. Some examples of these, and some control examples:

• Examples using articles currently mentioned in other sections above on this page:
  1. Talk header + project medicine ⟶ medical links included; e.g.: Talk:COVID-19 vaccine clinical research, Talk:Premenstrual syndrome Talk:Deployment of COVID-19 vaccines
  2. Talk header without project medicine ⟶ standard (non-medical) links; e.g.: Talk:Havana syndrome
  3. No talk header on page ⟶ no "find sources" links, regardless of project; e.g.: Talk:Clavicle fracture, Talk:Cerebral shunt, Talk:Peritoneal cavity

Note that in case #2, you can force the header to output medical links by adding the |domain= parameter: {{Talk header|domain=medical}}, and in case #3, you can add them separately (since there is no header) by adding template {{Medical sources notice}} to the page. One other case worth noting, is a page like Talk:Medtronic which belongs to the project, but for which the medical source links may be "too strict" for the page; in this case, you can force a different set of links using the |domain= parameter, like this: {{Talk header|domain=general}}, and the general "find sources" will be displayed instead of the medical links. For details, see template documentation.

You are the first to hear about this launch, and we could really use your feedback before making a more general announcement about it. Please try out the new "find sources" links that you find in the Talk page headers, and leave your feedback about bugs, feature requests, or general questions at Template talk:Find sources; or, if you prefer, below. Thanks, Mathglot (talk) 09:13, 8 November 2021 (UTC)

Interesting... probably a good idea. It comes to mind that this can "encourage" people to complex with MEDRS. Some thoughts:

1. Do we want to link to guideline search tools as well as papers. I don't know if they exist. I sometimes search NICE for stuff.
2. We could filter google scholar search to reviews (they have this feature). This might do another piece of nudges
3. For *some* topics (though not all). There are a few books that comprise most of the literature on the topic. This doesn't work too well for high-level articles that completely summarize a topic, but for others there are literally three reviews that do the bulk of the work. I wonder whether keeping track of *these* sources is
4. I heard that sci-hub implemented full text search recently.... Talpedia (talk) 11:37, 8 November 2021 (UTC)

Hi, Talpedia, and thanks for your comments. I edited your comment to add numbers to your points so I could respond in kind; hope you don't mind. (Feel free to revert, if you do.)

1. Not quite sure I understand. Are you talking about searching Wikipedia guideline pages?
2. Do you know what the url query string parameter is for google scholar reviews?
3. I'm aware that some niche topics have narrow sourcing, but not sure what you are suggesting we do in those cases. Can you give a concrete example?
4. My understanding is that Sci-Hub has copyright issues with their service, and I'm pretty sure Wikipedia's licensing requirements and U.S. copyright law would prevent us from using it. You could check with the legal department at WMF for guidance on this.

Thanks again for your feedback! Encouraging people to interact with and employ MEDRS is what this was all about. Mathglot (talk) 02:55, 9 November 2021 (UTC)

1. There are two types of reliable sources in WP:MEDRS review papers and guidelines about how to treat people.

   Examples include literature reviews or systematic reviews found in medical journals, specialist academic or professional books, and medical guidelines or position statements published by major health organizations.

   (WP:MEDDEF). This is an example of a medical guideline: https://www.nice.org.uk/guidance/ng136 . The often quote research and have the "truth" associated with being read by medical practictioners when deciding how to treat people, but also often have lower standards of scholarship compared to systematic reviews.Talpedia (talk) 08:27, 9 November 2021 (UTC)
2. Indeed, kind of a joke. But there is just a reality that sci-hub is being used by academics and PhD students for research even at universities in highly developed countries... Talpedia (talk) 08:36, 9 November 2021 (UTC)

   I believe that Sci-Hub ended up on the global spam blacklist for copyright violations. WhatamIdoing (talk) 06:58, 10 November 2021 (UTC)

1. I don't know of a global blacklist, but yes some academic publishers have been going around the world getting court injunctions that ISPs are applying. The implementation of this blacklist by ISPs is not necessarily consistent however. Sci-hub have plans to create legal and political challenges - I'm not particularly hopeful, but it might get more attention and force change. There does seem to be a move to different funding models - where academics pay paper directly for publishing their articles, which feels a bit crazy, but is perhaps representative of the transaction that is actually doing on (academics publish in papers for reputation) - but that doesn't deal with all of the old papers Talpedia (talk) 10:11, 11 November 2021 (UTC)

   This is our own global list, which is at m:Spam blacklist. It applies to all WMF-hosted wikis. WhatamIdoing (talk) 16:19, 11 November 2021 (UTC)

Thanks for sticking with this project, Mathglot. WhatamIdoing (talk) 22:09, 8 November 2021 (UTC)

@WhatamIdoing:, thanks. It's been a labor of love, both functionally, and technically. I just hope it serves the purpose, and of course, suggestions are always welcome. One thing I wondered about, is whether we should add a link for the Wikipedia Library back in. The "regular" find-sources links you get on non-medical pages has one, so maybe this one should, too. Or, if medical editors are used to having the Wikipedia Library bookmarked in their browser, maybe it isn't necessary? Need some feedback from regulars, on whether they'd like to see a WP Library link or not. Mathglot (talk) 03:00, 9 November 2021 (UTC)

Regarding the Wikipedia Library link, I should add that that link will soon become a lot more handy than it was; an upgrade is in the offing which will make clicking the link go straight to the search results, instead of making you go via the login page, retype your query, and submit it, the way it is now. For those who are interested, you can track this exciting development at Phabricator T294919. Mathglot (talk) 08:01, 9 November 2021 (UTC)

I'm looking forward to that feature.

I think that many qualifying editors don't know about Wikipedia:The Wikipedia Library or haven't found it useful for locating sources. Some of the recent improvements make it much more useful. WhatamIdoing (talk) 06:59, 10 November 2021 (UTC)

WhatamIdoing, this feature is now live. Scroll up or jump back to the top of this section, and try the 'WP Library' link now. I think this will really be a big help, and encourage more use of it. Mathglot (talk) 10:28, 11 November 2021 (UTC)

This looks great, it would be nice to highlight MEDRS sources rather than Newspapers etc when the template is shared on articles. For Guidelines, have you considered https://guidelines.ebmportal.com/ "The GIN international guideline library and registry of guidelines in development". I believe guideline groups have to pay to be shared in the this library, but it looks quite extensive. JenOttawa (talk) 13:28, 11 November 2021 (UTC)

@JenOttawa: On first reading, I assumed you were replying to Talpedia, but now I'm not so sure. If you were proposing some change to the find sources links to add something, can you elaborate? Thanks, Mathglot (talk) 04:31, 13 November 2021 (UTC)

Hear hear! Agreed this is exactly the kind of thing our wiki needs, and I'm glad you've made it a reality Mathglot :) — Shibbolethink ^{(♔ ♕)} 03:36, 14 November 2021 (UTC)

• I am very strongly opposed to this unfortunate development (see discussion below), and hope that this will be remedied ASAP by either incorporating text such as in the Ideal sources template, or removing this addition entirely. Why are we going backwards in making sure our editors can find and digest MEDRS? We are pushing inferior information to the top of talk pages, where editors will find it before they know what MEDRS is. SandyGeorgia (Talk) 16:10, 12 November 2021 (UTC)

medref template

Relatedly, the {{medref}} template suggests sources, but not of the right kind:

wpuld be good to improve this too. Alexbrn (talk) 20:23, 9 November 2021 (UTC)

@Alexbrn:, indeed. Was not aware of that template, I’ll get back to this when I’m not mobile; it should be an easy fix. Mathglot (talk) 00:54, 10 November 2021 (UTC)

I've responded at the template talk page (discussion). This needs feedback from the community, so members here are especially invited to have a look and share their thoughts. Mathglot (talk) 03:30, 10 November 2021 (UTC)

Agreed, taking the things in this top talk template and putting it into this section/page articlespace template would be a really good thing. — Shibbolethink ^{(♔ ♕)} 03:37, 14 November 2021 (UTC)

Mathglot set up a new sandboxed version that includes a short list of medical sources...

Discussion here. - Wikmoz (talk) 02:56, 22 November 2021 (UTC)

Alexbrn, Shibbolethink, SandyGeorgia, JenOttawa, Colin: This looks great to me but let us know if you have any objections to Mathglot pushing this live. - Wikmoz (talk) 22:23, 23 November 2021 (UTC)

I Think this is great :) No objections, these are the most high yield search engines — Shibbolethink ^{(♔ ♕)} 23:03, 23 November 2021 (UTC)

I do not see any problems, and it is a great improvement over what was there. Thanks! SandyGeorgia (Talk) 23:06, 23 November 2021 (UTC)

Certainly better than newspaper links. -- Colin°^Talk 10:53, 24 November 2021 (UTC)

Ideal sources is superior and is pushed down the page by this inferior template

This is a move in the wrong direction (while appreciating the well-intended effort). The very editors we have problems with will see the find medical sources in the header, and think that primary sources that they find in most of those links are adequate. I believe that the current {{Reliable sources for medical articles}} is more in line with what is needed, and am concerned that this new template will trump that one. To offset the potential detrimental effect, we would have to place the better "Ideal sources" template above this on talk, which would mean placing it above the talk header!!! For example, right now at Talk:Autism, editors will encounter this template with no further explanation at the top of the page, and are unlikely to scroll all the way down to where the "Ideal sources" template is displayed. The very editors we have problems with are most unlikely to scroll through every talk page template, and will see this first, and continue applying primary sources incorrectly.

I strongly encourage removing this from the talk header template, and instead incorporate the links into the superior template already at the Ideal sources template. My apologies for having (?apparently?} missed earlier discussion of this, and weighing in late, but please remove it now from talk headers until this can be sorted out. (Can someone please point me to the discussions where this was decided?) The "Ideal sources" wording is superior, and may now be lost in talk page clutter. {{Reliable sources for medical articles}}. SandyGeorgia (Talk) 16:00, 11 November 2021 (UTC)

Funny that the "ideal sources" mentions review articles but not med school textbooks and major guidelines, which are also MEDRS' ideal sources. WhatamIdoing (talk) 16:20, 11 November 2021 (UTC)

Then we should fix it. Now we have template proliferation, and a clearly inferior one pushed to the top of talk pages. I believe merging the new one into the old one would address that ??? SandyGeorgia (Talk) 21:32, 11 November 2021 (UTC)

I'm not sure that anybody reads the headers often enough for this to be a problem. This is more likely a hidden-in-plain-sight cheat code for experienced editors. WhatamIdoing (talk) 04:44, 12 November 2021 (UTC)

To be clear, this was an update to the {{findsources}} template, which has been on medical talk pages for a long time. It presents a standard interface that can be extended to other subject areas in the future. It didn't push down {{Reliable sources for medical articles}} (though perhaps this link list wraps to a second line on some screens where the prior one didn't). The template as of now includes no text regarding editorial policy or guidelines. If there's consensus that the "medical sources" label distacts from the more detailed WP:MEDRS text mentioned below, perhaps we can revert the label to "Find sources"? I think this was previously proposed by Mathglot and will help with line wrapping. In general, I believe that the templates can coexist or perhaps be merged in the future. It's possible that {{findsources}} could be expanded to display category-specific source guidelines (in this case WP:MEDRS and the explanation of ideal sources). However, I disagree that replacing general web search links with medical search links will result in the negative outcomes you describe. - Wikmoz (talk) 08:15, 12 November 2021 (UTC)

(edit conflict) Sandy, maybe some clarification is needed here, as nothing is "pushed down" the page by this change. Perhaps you misunderstood what this feature is about. Everything is in the same position as it was before. The only difference now, is that for those Wikipedia articles on medical topics which have a Talk header template (which appears to be a small minority of them), where readers previously saw the "standard" Find sources links (i.e., Google, NY Times, FENS (English newspaper sources), JSTOR, etc.) they will now see the new ones (Cochrane, Gale, OpenMD, PubMed, etc.) instead. The majority of medical pages do not have a Talk header template on them, so will not be affected.

For example: of those articles in Category:All WikiProject Medicine articles, I jumped to the letter 'S' and picked the first 20 articles beginning with 'S' and checked each one, and only two of them have a Talk header template; the other 18 do not, and thus are not affected by the new functionality. The remaining two articles do have a talk header template, and it is the same size now as previously, with all page elements in the same position as before. The only difference is, that now it shows the medical source links, whereas before it showed the standard links with New York Times, FENS, and all that. See Talk:Saaremaa virus for one of the two examples in the first 20 'S' articles in WikiProject Medicine where the Talk header template source links have changed over to the new style. That's too small (and not a random) sample of course to generalize from, but if that trend held for all articles in project Medicine, that would mean about 10% of them were affected by this change, hopefully in a positive direction.

To answer your first question: yes, it is easily undone. That would put back the NY Times, FENS, and other links, in place of the Cochrane-Gale-PubMed links, on the minority of medical articles which have a talk page header, without moving anything up or down on the page, and without affecting the majority which don't contain the template. Does this clarify at all what this new functionality does?

If you wanted to incorporate the links into the "ideal sources" template, that could of course be done, and is up to consensus; I have no opinion on that. That wouldn't change positioning of the Talk header template on pages where it appears, though. It has always been possible to suppress the "find sources" links so nobody sees them on a given page by means of a template parameter. It is also possible for a user to set a preference so that you don't see Talk page header templates at all, regardless what parameters it has, and that will affect just the user in question, and will move the "ideal sources" closer to the top of the page, for those pages which have both templates on them. That would be an "opt-in" type change to common.css, which very few users are likely to take advantage of, and almost no new users. Mathglot (talk) 09:06, 12 November 2021 (UTC)

See Talk:Autism. Readers will now encounter this (incomplete) information before they encounter the Ideal sources information, unless we breach talk page layout guidelines and put the ideal sources first. This needs urgent fixing, and will lead to problems. We can’t go about suppressing this incomplete info on every talk page. This has created an urgent problem everywhere. I disagree with WAID that new editors will not be misled; I’ve seen it too may times with my own eyes. WE should not have to suppress this (faulty, incomplete) information. If you can’t fix the bigger problem by incorporating the Ideal sources wording, then the default should be for this text to be off in the Talk header, requiring us to activate it if we want it. For now, this forces us to deal with faulty incomplete information being at the top of too many medical articles. SandyGeorgia (Talk) 16:06, 12 November 2021 (UTC)

In the meantime, please let me know how I can suppress this information from the talk headers on pages I watch. In Dummies101 language. SandyGeorgia (Talk) 16:11, 12 November 2021 (UTC)

Sandy, I changed the links in Talk:Autism (diff ) so it's back the way it was before with the NY Times and other links by following the instructions at the last bullet below. Do you find the header better this way? Or are you saying you want all the "find sources" links gone entirely so the first links one encounters are the "ideal sources" lower down? The template documentation explains how to hide or change various features:

• To suppress only the "find sources" links and keep the rest of the template as is, see: Template:Talk header#Hiding the links
• To suppress the "header template" entirely, but only for you, see: Template:Talk header#Hiding the template. (To suppress it for everyone, remove the line in the wikicode of the page that looks like this: {{Talk header|...}}.)
• To change the links shown in one page from the new, medical domain links back to the old ones, see Template:Talk header#Search domain override.

The questions about "Ideal sources" wording or position, I see as a separate issue from this one, which was discussed in August I believe; I'll find the links and post them later for you. Anything *can* be done, it's a matter of finding out what the consensus is, and doing it. If the consensus is to go back to the way it was before with respect to the "find sources" links, then that can be done; and if the consensus is to consolidate the wording into either the "ideal sources" or the "find sources" sections, then that can be done; and if the consensus is to switch the position of the "ideal sources" links so they're on top, then that can be done, too. Meanwhile, there's always WP:BE BOLD; I've boldly moved the "ideal sources" links at Talk:Saaremaa virus to the top of the page, *above* the Talk header. How does this seem to you? Would it address your issue, if the "ideal sources" appeared on top, as in this example? Mathglot (talk) 19:27, 12 November 2021 (UTC)

I am iPAd typing so can’t go searching for a link, but the adjustment made at Talk:Autism does not address the problem. We are still encouraging new or inexperienced editors to misinterpret MEDRS in the very first thing they see on talk. SandyGeorgia (Talk) 20:36, 12 November 2021 (UTC)

Please remove this new content from the talk header ASAP, and re-add it only after these problems are sorted— we cannot and should not be making such broad changes across so many articles. SandyGeorgia (Talk) 20:38, 12 November 2021 (UTC)

Sandy, I'm struggling to keep up with your concerns, and understand them. You said:

the adjustment made at Talk:Autism does not address the problem. We are still encouraging new or inexperienced editors to misinterpret MEDRS in the very first thing they see on talk.

Why doesn't it address the problem? Talk:Autism is back to exactly the way it was a couple of days ago; isn't that the way you wanted it? And what do you mean by the "very first thing they see on talk"?

Regarding "ideal sources" and where they are positioned, can you respond to my question above about whether you like the way Talk:Saaremaa virus is now organized, with the "ideal sources" at the top? Is this what you prefer?

As far as doing anything ASAP, with all due respect, this should be based on consensus; you are one editor, and if your view represents consensus, of course it will be removed, but as I said, I'm still struggling to understand what it is you want exactly. The change was discussed previously at the Village Pump, and advertised here, and there was sufficient approval to go ahead with the changes. All I understand for sure, is that you want everything put back the way it was, without understanding exactly what it is you don't like, which may be fixable. You want "ideal sources" always at the top of the Talk page for medical articles? That is eminently doable, although it is independent of this change. Is that what you want? Mathglot (talk) 22:35, 12 November 2021 (UTC)

Found all of the discussions myself, and the discussion here on WT:MED that involved pretty much … no one. Very disappointed that such a major change can be made with such far-reaching consequences based on such little input. The generic “find medical sources” (with no link to WP:MEDRS) is unlikely to produce a good outcome on controversial medical articles, where we are now telling new editors that everything in those links might be useful, and providing no context indicating that, in the majority of cases, they may not be. I hope this will be rectified by at minimum providing a link to MEDRS somewhere in there. Meanwhile, this template is misleading on medical articles, and has been since this change was first initiated in November 2020. [1] SandyGeorgia (Talk) 03:08, 13 November 2021 (UTC)

Just wanted to add my 2 cents to say I completely disagree and would be very unhappy to see this talk banner go away. I think it's an improvement to the project and exactly where editors should be looking for high quality reliable sources. — Shibbolethink ^{(♔ ♕)} 03:39, 14 November 2021 (UTC)

No question that systematic reviews and guidelines are the ideal sources. Editors should be strongly encouraged to gain a basic understanding of the hierarchy of medical evidence and learn to check those sources to validate new findings.

I do question whether the talk page heading sequence makes that much of a difference in visibility (the bottom position like this closest to the ToC may actually be more visible than anything in the middle). I also question the claim that improving access to medical sources is a negative (regardless of the page sequence).

• If the vertical position of a header dramatically impacts its usage then WP:TALKLEAD should be immediately revised to move {{Reliable sources for medical articles}} up 10 positions to directly below {{talkheader}}, which would also make it immediately adjacent to {{findsources}}. So that's a possible solution pending consensus and presumably the higher priority.
• If the "Find medical sources" label is the issue then changing the label to "Find sources" is another easy solution pending consensus.
• If ensuring that category guidelines (in this case WP:MEDRS and text directing users to prioritize ideal sources) appear above the {{findsources}} links then the best solution may be to modify the {{findsources}} to accept a preamble text field where a block of HTML can be displayed above the links.

All that said, just in terms of urgency and real world impact, we have no usage data so it's hard to tell what percentage of editors actually read or use any of the talk headers. I suspect that talk messages are helpful to experienced editors (particularly topic regulars). I also suspect that newcomers who post misinformation or poorly sourced information do so BEFORE visiting a topic's talk page, BEFORE reviewing the headers, and BEFORE determining that they need to respect any message in the headers. - Wikmoz (talk) 21:08, 12 November 2021 (UTC)

Indeed; there's such a thing as banner blindness, and we all have it. If I asked you to close your eyes and tell me what's in the top left portion of the Talk header box, could you do it?

Answer to banner blindness quiz.
Put new text under old text. Click here to start a new topic. Sign your posts by typing four tildes (~~~~). New to Wikipedia? Welcome! Learn to edit; get help.

If it becomes important enough, WMF personnel have access to logs that can show how often these links are actually being used. I suspect we are worrying about something that gets used hardly at all; which doesn't mean we shouldn't find the best way of presenting the headers, the links, and everything else on the page, but it does mean, I think, that we needn't have such a sense of urgency about this, and that a calm discussion will lead to the best result. Mathglot (talk) 23:39, 12 November 2021 (UTC)

Source of problem identified … in a page very few editors follow, here. SandyGeorgia (Talk) 02:52, 13 November 2021 (UTC)

That was a proposal to remove some wording from the Template documentation that currently discourages users from adding Talk header templates to empty pages, and instead, "[to allow] users to add the template to any talkpage, including new ones". I don't see a connection between that discussion and the issue we are talking about; they seem unrelated to me. What connection do you see? Mathglot (talk) 03:06, 13 November 2021 (UTC)

I don’t think we should be using the template anywhere now; it’s a misleading mess (and has been since Nov 2020). The default should be that none of that misleading info is included in a template that is widely used, and if someone wants find sources added, they can turn it on-- no one is going to turn on misleading info about sourcing on medical talk pages, and yet it was installed automatically anyway. The change I want to see is that a) regular editors should not be required to learn complex syntax to implement a basic talk header, and b) a link to MEDRS needs to be somewhere in the “find medical sources” line, which is now misleading, as most of what inexperienced editors will find there may not be usable. And finally, all template editors should be aware that broad changes like this should be broadly advertised; regular editors find out after the damage is implemented.

Also, now that I have figured out where the problem originated, I don't think we should hold this discussion in two places; it looks like Template talk:Talk header is where this problem originated and was discussed among very few editors (similar to the very few who discussed it here). IF that kind of misinfo is to stand across all medical talk pages, we need a community-wide RFC to address the misleading MEDRS implications. SandyGeorgia (Talk) 03:18, 13 November 2021 (UTC)

Just for reference, the proposal for this update was posted here with an invitation for WikiProject Medicine feedback posted here. - Wikmoz (talk) 03:22, 13 November 2021 (UTC)

Thank you for that; most appreciated. (I took most of the middle part of the year off because of some other Wikipedia dysfunction, but had I been regularly editing, I doubt I would have seen those anyway; this kind of dysfunction will encourage me to take the rest of this year off, as it defeats everything we have worked for for years in one backchannel edit.). I see the same few editors weighing in there, so obviously it was not a very broad discussion. At any rate, that is water under the bridge, and I hope you all will fix the misleading implications so a community-wide RFC is not needed. A link to MEDRS should not be that difficult, is it ? Most of what is listed as “medical sources” will produce sources that won’t be useful, and new editors will find themselves reverted after following talk instructions. SandyGeorgia (Talk) 03:27, 13 November 2021 (UTC)

Sandy, Wikmoz beat me to it; I was looking for the links at the same time. It was first added to WT:MED at 01:55, 10 August 2021, and remained until it was archived at 05:09, 29 August 2021. It notified WP:MED editors of a Village pump proposal discussion started on 12 August 2021 which remained open until 20 September 2021 when it was archived to WP:VPR/Archive 184. There was light participation, but WP:VPR is a widely read, centralized discussion forum, and few objections were noted. Light participation can also mean nihil obstat; not everyone comes in to be part of the cheering section when they like a proposal, but if there are any doubters or nay-sayers, you can be sure you'll hear from them quickly, and many proposals don't make it past the proposal stage; so I feel that this one has consensus.

Consensus can change, and I could see you leading the charge to change it, and that's fine. But I don't feel that I can unilaterally overturn the result of that consensus on my own at this point simply by removing what has been added; there would have to be at least as much consensus to remove it, as there was to add it in the first place. Whether it is a community-wide RFC that decides that or just plain discussion on a centralized forum or something else, I just wouldn't feel right marching in and undoing it without some backing. The only thing I would say, is that if you do decide to do an Rfc, please pay special attention to how you word the Rfc question; it seems to me from the discussion above, that four or five inter-related but not identical topics have been raised, and if one tries to solve everything at once in one Rfc, the choices proliferate, and it might become confusing; better to stick to a yes-no question, if possible. Certainly Rfc questions like, "Do you support removing the recent dynamic find-sources links that added medical sources to Talk header templates", or "Do you support removing all find-sources links from Talk header templates?" are well-defined questions, if that's what you're going for. So is, "Do you support moving the {{Reliable sources for medical articles}} template to the top of the Talk page (or, '...above the Talk header template') on Talk pages which contain both templates?" Or even, taking the Talk header template to WP:Afd to delete it entirely, if that's what you prefer.

I'm not sure if Sdkb is lurking, but we haven't heard from them so far. Maybe they'll be better at responding to your questions and issues than I have been. Cheers, Mathglot (talk) 04:23, 13 November 2021 (UTC)

Hi! I'm present, but I haven't read all of the above. Is there something particular you'd like my thoughts on? {{u|Sdkb}} ^talk 06:10, 13 November 2021 (UTC)

@Sdkb:, well, you might have to read from the top to get the full flavor of it I guess, but I just feel I haven't been effective in responding to Sandy's concerns in a way that she and the topic deserves, and I thought maybe you might be able to do better, if you could. Mathglot (talk) 07:24, 13 November 2021 (UTC)

Sdkb, it’s not complicated. The “find medical sources” wording followed by a list of places to look is going to mislead inexperienced editors and result in LOTS of sources that are not adequate for sourcing medical content, per WP:MEDRS. See counterexample at Template:Reliable sources for medical articles. A link to WP:MEDRS somewhere in the new wording would help. Now, the talkheader (placed at the top of article talk pages) will take precedence. I am sorry to find us here, with the respect I have for so many of you, but this is an epic fail that requires a speedy fix, if not an RFC. There is CLEARLY no broad consensus for this addition, as the participation in those discussions was so light, and mostly the same few editors. For such a simple problem, we should not have to sort through walls of opposition technical text: fix it please. And, rather than waste my time trying to “lead the charge” to get such an obvious epic fail fixed, I am more likely to just unwatch all medical content again, as this is SO discouraging in how one template edit undoes 15 years of work towards improving medical content. This is a much bigger issue than the socking affecting medical content that led me to stop editing most of this year. There is no reason to work to improve medical content when so much work can be undone in one edit to a template spread across all medical content, and misleading the very editors we need to reach — the new and inexperienced. SandyGeorgia (Talk) 16:05, 13 November 2021 (UTC)

I'm not enough of a medical editor to wade into this further, sorry, except to say that there are unfortunately no shortcuts to consensus. So if you think the previous RfC erred, starting a new one would be your main option. {{u|Sdkb}} ^talk 17:08, 13 November 2021 (UTC)

Sdkb could any of you point me to where in the very limited consensus this addition relied on we find any consensus preventing you from adding a link to WP:MEDRS, or where the actual content of the exact addition gained consensus, and also address why editors unfamiliar with medical editing are making a template edit that broadly affects every medical article? SandyGeorgia (Talk) 17:26, 13 November 2021 (UTC)

There is nothing stopping us from adding a link to WP:MEDRS. How do you want it to read? We could, for example, link the already existing words 'find medical sources' so it would become [[WP:MEDRS|find medical sources]] in the wikicode, resulting in find medical sources, for example. Or, we could add a new link, first in the series, last, or somewhere in the middle; we'd need wording for the link text. Maybe that could be via a tooltip (although that won't help mobile users), like this: find medical sources. Mathglot (talk) 18:09, 13 November 2021 (UTC)

It that is an option, a) I am greatly relieved, but b) think Colin is our best wordsmith, so we should make that a separate discussion. Thanks, Mathglot ... this is hopeful. SandyGeorgia (Talk) 19:13, 13 November 2021 (UTC)

I know Mathglot has been working on a flexible solution for linking WP:MEDRS in the label. For the moment, I've added it to the source list as the first link. - Wikmoz (talk) 16:30, 14 November 2021 (UTC)

Wikmoz, Most helpful! Is it possible to add any plain text along with that, or is that asking too much of the software? If it is possible, we could come up with three or four words that amount to something akin to “read me first”. SandyGeorgia (Talk) 16:54, 14 November 2021 (UTC)

Yes. It's possible to add text in parens after a link if there's consensus to do so. Ideally, we'd be able to keep it to two words (editor notice, read first, ideal sources, etc.). - Wikmoz (talk) 17:03, 14 November 2021 (UTC)

Continued below at #Link to MEDRS. SandyGeorgia (Talk) 17:15, 14 November 2021 (UTC)

I tried this out at Tuberous sclerosis (added {{Talk header}} to top of talk page and previewed the result). The results are a bit of a mixed bag. The Cochrane search at TSC did offer two Cochrane Reviews which I guess is a good result, but the equivalent for Marie Curie did not find any reviews so the tab on the web page defaulted to Clinical Trials which would make most of us weep. And so on with the other links. Perhaps what is needed is a help page linked to from that little "Find medical sources" section. It could have sections each for Cochrane, JSTOR, Google Scholar, PubMed, etc. That could then guide editors as to which tabs on the Cochrane results are worth using, etc. I think it is possible the links could be very useful but we can't control the search results as much as we'd like, so editors will need some training as to how best to use each search. I think that training could be better done in a project help page than in any official guideline page. -- Colin°^Talk 10:44, 13 November 2021 (UTC)

Colin, that seems like a good idea. Do you think such a page should be tied directly to the dozen or so links produced by {{find medical sources}}, or be a superset of it that would contain those at a minimum, plus whatever other popular sources users might find or use, that we don't have room for in the link set? One advantage of the latter approach, is that we could include a section waving off users from using sources that were known *not* to provide very good results for medical articles, or just have a caveat in the description for the link, explaining what's good about it, and what to watch out for. Mathglot (talk) 18:16, 13 November 2021 (UTC)

As part of this, we might examine some of my analysis below as to whether we really want to include every one of those links ... my analysis is still too preliminary to say ... SandyGeorgia (Talk) 19:14, 13 November 2021 (UTC)

Sandy, the choice of links is easily modifiable. (Per JenOttawa's request above, the Guidelines International Network (GIN) has been added; it can just as easily be modified or removed.) I can't comment on the quality of medical content, or the quality of these specialized searchable medical collections generally; I just wanted to say that all of the links generate a search results page using the article title as the query terms. Just like Google or any other search engine, the search results are not curated, they are algorithmically determined and will find any document in their respective database collection that matches the query terms, sorted by whatever their relevance algorithm is, and like Google, such a query will turn up hoaxes, pseudoscience, or controversial claims along with the good stuff, if their database collection includes documents which contain the search terms (i.e., our article title). They will also happily turn up results older than five years (or older than fifty years) unless their relevance algorithm favors recent articles.

To the extent that we would warn users performing a google book or scholar search that they need to check the date and the quality of the publication, the same caveats would apply to any of the other specialized database search engines such as Cochrane and the others. This is not something that's easy even for experienced editors to do, let alone newbies, and I think Colin's idea above of a page explaining individual medical sources (perhaps informed by, or with links to discussions at WP:RSN in some cases?) is a great place to start.

As far as analysis, I'd just add that paucity of results or none at all can convey information, too; the fact that Cochrane has only two review articles for "Morgellons" doesn't necessarily mean that Cochrane results are systematically poor; conversely, the fact that Springer has 18 for ROGD, including letters to the editor, methodology critiques, and opinions in educational journals, doesn't mean that Springer endorses those results and is useless for everything, but that this is what little is available for those query terms, and maybe none of those results qualify for MEDRS. So I would just say, when evaluating collections like Cochrane, PubMed, Wiley, and so on, think about what your measurement yardstick is a priori. No search engine will return results that are 100% MEDRS-compliant, and there's no getting around the work of learning how to separate the wheat from the chaff in such result pages. A document like that conceived by Colin, could address those issues. Mathglot (talk) 20:52, 13 November 2021 (UTC)

Thanks, Mathglot … I don’t disagree with any of that, although it does begin to look like we are just creating the equivalent of a not-very-useful link farm. Again, my concern is more that we just fix the wording with, at minimum, a link to MEDRS. I doubt that the editors we need to reach will read much anyway, but we need some sort of advisory or dislaimer. I am concerned that we need to fix this as soon as we can, and fear that Colin’s idea will take longer to put together. Can’t we just, for now, link WP:MEDRS to the “find medical sources” wording? That is at least a temporary something … SandyGeorgia (Talk) 21:00, 13 November 2021 (UTC)

Sandy, yes, I tried doing that already, and it worked for {{find medical sources}} in isolation, but for some reason, when invoked by Template:Talk header it broke badly, so I had to back it out (history). I was going for something like this:

{{#invoke:Find sources|Find sources medical/sandbox|influenza vaccine}} Note: example commented out; subsequent sandbox changes broke it; by Mathglot (talk) 21:06, 14 November 2021 (UTC)

It may have something to do with the Lua code or the config; in any case, I'll have to ask for assistance with that at the module.

As for Colin's idea, I wrote up a scaffold in Draft space, so Colin could concentrate just on creating the medical content and expertise that I can't help with, and not have to deal with the framework of the page. Colin, you can find it at Draft:Wikipedia:Using medical database collections; feel free to improve it, rewrite it, or ignore it. Everything in there that isn't obvious boilerplate, was copied from existing project pages (like MEDRS) or template instructions, so I haven't added any original material on my own, other than one background paragraph on general principles of search (which I'm expert in) and some connecting words and segues and that sort of thing. Mathglot (talk) 01:16, 14 November 2021 (UTC)

Wrt searching sources, I'm a bit out of practice compared to others here, and many of these searches are new to me, so I'm not really the best person to guide anyone about them. You know that "a little knowledge is a dangerous thing" and I wonder if those search links make it seem very easy to find sources without really addressing the hard work of selecting from the results or refining one's search terms. I also wonder the effect of having hundreds of links to some publishers but not others. Sure, it is handy to point potential editors towards useful search engines, but are they really needing the extra hand-holding of having someone type in the search keyword. Are there people flitting from article to article desperately needing to urgently make some random improvement, and so time-poor that typing in "epilepsy" into a box was such a killer they gave up? If, as Sandy fears, we end up at WT:MED dealing with the fallout of "Well, you told me these were suitable medical sources... and now it turns out they weren't", then is it a plus? I wonder if some college course will direct student editors towards clicking on the links and adding something citing the first few results. Also, why on earth is Cochrane first? I'd have though the pubmed search for reviews was the primary one to steer editors towards. -- Colin°^Talk 10:23, 14 November 2021 (UTC)

Samples

Colin’s example of Tuberous sclerosis reveals a “mixed bag” for even an article that is not a highly edited or controversial article (eg of the type where we most need to point new, inexperienced and advocacy editors towards good sources). Examining some of those reveals how this template will exacerbate the problems in articles that are highly hit by advocacy editing, by misleading those new, inexperienced and advocacy editors towards unusable sources. SandyGeorgia (Talk) 17:30, 13 November 2021 (UTC)

TEMPLATE SAMPLE RESULTS: Evaluation of template results for 6 medical topics

Morgellons - Talk:Morgellons Morgellons: completely unhelpful, and exacerbates the very problems now encountered in this controversial topic Cochrane: nothing. DOAJ: the first three returns (and most of what it produces) are the very sources we disallow on this topic, that are constantly added by advocates, fourth is a review Gale: similar to DOAJ, and most breach WP:MEDDATE Google Scholar/Books: very first hit is published by the advocacy group that creates havoc in this topic JSTOR: one result which is gibberish; if I don't sort it, new editors are unlikely to. OpenMD: first result is to info at NIH which would not be used in this article anyway, and sends the reader to Mayo, not a MEDRS-source; second hit is Mayo; there are usable sources after those first two, will inexperienced readers know the difference? PubMed: first hit is MEDRS-compliant; second is very old review (not WP:MEDDATE); third result is a usable review; then we descend again into Middleveen, not usable. ScienceDirect: first result is usable, then quick degeneration to unusable case reports and correspondence Springer: grossly unhelpful per a combination of all of the above Trip: grossly unhelpful per a combination of all of the above Wiley: first fails MEDDATE (this is a controversial evolving topic); second is usable bot old, third and fourth fail MEDDATE ... WP Library: What. The. Heck ??? If I am not going to negotiate this page, will new and inexperienced editors? SandyGeorgia (Talk) 17:17, 13 November 2021 (UTC) Chronic Lyme disease - Talk:Chronic Lyme disease Chronic Lyme disease: mostly unusable but highly variable, hence confusing Cochrane: mostly very dated clinical trials, something useful shows on 6th entry, but not anywhere near the quality we'd expect (3 cases report and review) DOAJ: old to very old or case reports are the first hits; digging in deeper reveals very specific sub-topics, unlikely to be accessed by new or inexperienced advocacy editors Gale: first is a case report; second is very old; third is advocacy; fourth is old but could be useful (most likely newer sources are better) Google Scholar/Books: first is very old; second is old but might be useful; third is very old; fourth is very old JSTOR: nothing OpenMD: first is old but could be useful; second is NIH which can be useful, although underlying literature is better; third is CDC which can be useful, underlying literature is better; fourth is very old PubMed: first is a useful review but somewhat old; second could be useful, is not English (Dutch); third is advocacy; fourth, not sure; fifth is an unusable editorial ScienceDirect: first is case report, second is news, third is a survey, fourth is a recent review Springer: first is old; second is a slightly dated review; third is a potentially useful recent review; fourth is an unusable case series Trip: useless, a bunch of primary research Wiley: first four are so dated as to be useless, didn't even look into whether they are reviews WP Library: another What the heck. Alexbrn could you look at these two I have posted, as you may know these sources better? SandyGeorgia (Talk) 17:18, 13 November 2021 (UTC) PANDAS- Talk:PANDAS PANDAS A topic I know well, noting that the alternate Template:Reliable sources for medical articles leads us straight to very usable, only recent, secondary reviews. Mostly useless returns, definitely worse than the Reliable sources template returns. Cochrane: trials and extremely old dated info DOAJ: difficult because we get Panda bears mixed in with PANDAS, first unusable; second is a dated review useful only at a highly experienced level (not likely to be used by advocacy editors); third, fourth, fifth and sixth are bears, any sane person would give up. Gale: in this order of hits, bears, bears, primary study, bears, dated, bears, dated ... Google Scholar/Books: in order of hits, bears, very dated review, unrelated, unrelated ... JSTOR: bears, unrelated, bears, bears, unrelated ... nothing useful OpenMD: first hit is WebMD, useful, but underlying literature is better; second hit is NIH, which could be useful, but underlying literature is much better because Susan Swedo is not an independent source (she invented the controversial diagnosis); bears; primary study; news; bears; dated; then UptoDate, which is reliable secondary but requires subscription ... so well in to the hits and finding no recent MEDRS-compliant review PubMed: as with the MEDRS-compliant search on Pubmed in Template:Reliable sources for medical articles, this unrestricted search on Pubmed also returns mostly recent secondary reviews. ScienceDirect: first six hits are bears, stopped looking Springer: in order of hits non-English; unrelated; unrelated; unrelated; bears ... stopped looking Trip: first hit is review but unrelated; secondary review for Norwegian application; unrelated; unrelated; unrelated; primary study Wiley: extremely dated Susan Swedo; case report; very dated; very dated; dated ... WP Library, not going there SandyGeorgia (Talk) 18:48, 13 November 2021 (UTC) Sandy, thank you so much for taking the time to do this. It's incredibly helpful in clarifying your objections. I will take a closer look at your findings later today. - Wikmoz (talk) 19:45, 13 November 2021 (UTC) Wikmoz and thank you as well for being willing to have a look. I admit to being quite perturbed early on at what seemed to be nonchalance towards a very serious problem. I hope we can quickly just fix the wording, and put an end to this problem. I was planning to work up several more examples, but would be very happy to hear they may not be needed ... please let me know. Regards, SandyGeorgia (Talk) 19:51, 13 November 2021 (UTC) Made a minor in-line edit above to add links to the search results in question (and talk page with the corresponding ideal sources links). No need for additional examples. I think this gives us a lot to work with. - Wikmoz (talk) 20:03, 13 November 2021 (UTC) Thanks! SandyGeorgia (Talk) 20:39, 13 November 2021 (UTC) Tuberous sclerosis Cochrane: The two systematic reviews are five and eight years old, and both likely to be superseded elsewhere. I hit a problem with my Firefox browser. After searching for an author's contribs, now that's all I ever get. The links no longer work. DOAJ: Case reports, letter to the editor, non-English papers, basic molecular science. The website has no way to limit searches to review, etc. So really this is just wasting my time and why would I use it rather than PubMed? Gale: As above. And no way to filter search. Useless. Google Scholar: Old reviews and a very old book. Even tightening the result dates doesn't produce results that are better than PubMed. Google Books: Possibly some useful results, but I'm sceptical. I've too often seen Google Books being cited not because those books or chapters were great sources, but because that's what turned up. JSTOR: First result was an irrelevant article that just happened to mention "tuberous sclerosis" and yet the results were apparently sorted by relevance? The second, from 1918, was "General Aspects of the Brain Anatomy of the Feeble-Minded". Hmm. None of the other results were useful. OpenMD: Mostly tertiary websites. Potentially has some use in guiding article writer about what to cover and present in lay terms, but not ideal as sources. PubMed: As we expect, filtered to select reviews. A mix of reviews though many were not free to read. But it can sometimes be better to search for the text in titles and then order by date. ScienceDirect: First result was this which is good. I was puzzled that PubMed didn't find those recommendations. I'm thinking they didn't fit into the search filter for reviews. Not sure what such guidelines are classed as, though the ScienceDirect page classed it (wrongly?) as a "research article". ScienceDirect does let me filter for reviews, practice guidelines, etc. So this could be useful to see if it produces results ordered/grouped differently to PubMed, though it is limited to one big publisher. Springer: The initial results were all just brief mentions in some reference work, which wasn't useful. The search for "articles" didn't offer any more refinement. So I don't see what this is buying when I could get these searched better by PubMed and not be restricted to one publisher. Trip: Can we get it to filter by "All secondary evidence" by default? It returned some useful UK guidelines. Wiley: First few results were book chapters, which you obviously need the book for. The article results were not filtered or ordered in any useful way. And restricted to one publisher. GIN: nothing WP Library: Very mixed results, though the articles could be read. There may be some useful stuff if you are prepared to sift through this, though there is no way to filter for article type. Tourette syndrome - Talk:Tourette syndrome Tourette syndrome: an FA I wrote so know the sources well Cochrane: unuseful. Four entries only. Extremely dated; extremely dated; dated but somewhat useful, although unlikely to be used; somewhat dated and unlikely to be used DOAJ: very useful. First entry is March 2021 extremely helpful, and I need to look into using it. But Pubmed would have produced it as well. Second, third, fourth not usable. Fifth can be used (2020 update on DBS, but Pubmed would produce it as well). Gale: useless. Primary, primary, news, primary, primary ... JSTOR: completely useless-- have a look, perhaps something is wrong in the settings. OpenMD: unhelpful. Their basic definition at the top of the page lacks nuance ("uncontrollable". First source is Brittanica, unusable. Second source is NIH, already used. Third is Mayo, unusable, although relieved to see they have corrected their "uncontrollable" wording. Fourth is MedlinePlus, which is child-like. PubMed: not overly helpful, but not misleading either, but TS is already at FA level. First result is 2015. Second is 2018, already in the article. Third result is Turkish 2016, not helpful. Fourth, nothing we would use (2018 research highlights). Fifth, already in the article. ScienceDirect: useless Primary, primary, primary. Fourth is "in press" and the English is so poor I would advocate against its use. Fifth is a recent review from the B-team that wouldn't be used, as we have superior sources. Springer: variable. Mostly dated, but some of it could be useful if the article was not already at FA level (we should look at other articles for this case). Trip: somewhat helpful. The first entry is a review of treatment for Swedish; it looks too simplified and general to be of any use. Second is BMJ 2018 best practice, very brief; would be useful if article were not already at FA level. Third is Cochrane review of Botox. Fourth is USA guidelines. Fifth is highly technical. Wiley: useless, ugh, dated, have a look. GIN: nothing found ??? Still working, will next do a high-view article that is not already at FA level. SandyGeorgia (Talk) 19:31, 14 November 2021 (UTC) I think you may want to reassess your ScienceDirect inquiry, using the filters I added to that search button.(e.g. click it again) It excludes primary articles. — Shibbolethink (♔ ♕) 19:33, 14 November 2021 (UTC) Shibbolethink Did you just make that change? Because I viewed it within the last 15 or 20 minutes, and now it is good-- returning only recent reviews. Struck above, SandyGeorgia (Talk) 19:37, 14 November 2021 (UTC) Yes, just about five minutes ago! Glad to hear it helped :) — Shibbolethink (♔ ♕) 19:38, 14 November 2021 (UTC) Parkinson's disease - Talk:Parkinson's disease Parkinson's disease: a former FA that desperately needs a rewrite to good sources, top 30 most-viewed among medical content, and a topic I know well enough to opine on sources but not well enough to restore the article to FA level. This article needs a complete rewrite to newer sources. PubMed: bingo (is it possible to sort newest to oldest)? Cochrane: very dated, unlikely to be of any use at all. Perhaps this could be fixed by finding a way to sort by newest first? But for an article that needs a top-to-bottom rewrite, Cochrane tends to be very specific, and here we need broader reviews. DOAJ: not helpful. 1, dated and very technical. 2, dated, and not broad (limited topic). 3, same. 4 is case reports. 5, dated and very specific. Gale: variable. Primary, primary animal model, recent review, primary, recent review. JSTOR: not useful. 1, dated not helpful. 2, very limited in scope, but could be used for a sentence or two. 3, what ? 4, dated, not useful at all. 5, completely dated. OpenMD: unhelpful, all of these are pretty much commercial or non-MEDRS, have a look, not sources to which the article should be rewritten. ScienceDirect: bingo, with the new settings, this is now returning recent reviews. Springer: not too helpful. 2001 book, 1983 book, 1977 book, and by the fourth entry we get a 2020 book. Considering there are plenty of recent reviews, why would we resort to a 2020 book, which many won't have access to and where contents (written probably in 2019) may be already replaced by newer knowledge? Trip: very narrow utility, could be used to cite a few sentences. 1, very limited. 2, UK guidelines. 3, very limited UK guidance (?). 4, European guidance on one drug? 5, Swiss use of physiotherapy. Wiley: not useful, everything is way too dated and specific or primary, could have been replaced by newer reviews GIN: useless, nothing found, odd for such a common condition.

Let me know if I should do another or some different type of condition-- hoping this is enough for discussion purposes, SandyGeorgia (Talk) 20:17, 14 November 2021 (UTC)

Discussion of samples

It definitely looks like there are areas where we can adjust the sources and search parameters to improve the results. I also agree that it would be good to add a WP:MEDRS link and related text.

However, I don't think there's a silver bullet to consistently produce perfect results. The Trip results you flagged as useless are the same ones produced by {{Reliable sources for medical articles}}. The York CRDWeb link doesn't default to recent studies. The PubMed link isn't limited to high quality (Index Medicus) journals. The CDC search for "Morgellons" includes blog entries (notably, "Preparedness 101: Zombie Apocalypse").

Two key points for discussion:

• MEDRS-ideal vs. MEDRS-compliant: MEDRS strongly favors recent systematic reviews, guidelines, and text books from highly reputable publishers and reputable government agencies. For lack of a better term, I'll refer to these as "MEDRS-ideal sources". While discouraged, MEDRS does not prohibit the use of older reviews or primary sources--all the way down to in-vitro studies. Rather it provides guidance on how to prioritize available research and when primary studies can be used. Indeed, primary RCTs are used extensively in Wikipedia medical content. A recent large-scale triple blinded RCT published in a leading journal certainly passes MEDRS and could well supersede an earlier systematic review based on more limited data.

• Context: The degree to which editors need to actively police referenced material according to MEDRS recommendations (not the requirements) varies from topic to topic based on the volume of available research and the level of interest from advocacy groups in advancing a certain perspective.

Considering the above, my questions are:

• To what degree do we restrict all users from considering visiting sources other than MEDRS-ideal sources? WebMD and Mayo Clinic (which you identified as not a MEDRS source), may be a great resource for an editor doing a sanity check to verify that a Wikipedia article covers all the basics and that nothing in the article seems to go dramatically against current consensus... and accomplish this without hitting paywalls or spending hours interpreting the findings of systematic reviews on each question.

• Should we focus on evaluating the utility of source suggestions as they apply broadly to most medical topics or in the context of controversial topics?

We definitely need input from more editors. From my perspective, we should guide users to searches that produce very good results, even if they're not all ideal. The vast majority of results produced by the suggested sources are MEDRS-compliant (if not MEDRS-ideal) and better than those found on Google, mainstream media, Facebook, and random health blogs. Some supporting text could point users to systematic reviews and caution users that WP:MEDRS guidelines supersede the mere presence of a study in any given result set.

Several times in the past few years and particularly on contentious topics, I've encountered systematic reviews published in foreign or low impact factor journals that drew highly questionable conclusions. In cases like these, it's helpful at least be aware of recent well-structured RCTs that contradict the review.

I'll post some possible suggestions below so the philosophy discussion and fixes have separate threads. Wikmoz (talk) 00:31, 14 November 2021 (UTC)

Suggestions:

• PubMed: Already limited to reviews but would benefit from having a 10 year date filter added (&filter=datesearch.y_10). ( Done)

• Trip: Defaults to a quality sort with secondary evidence placed at the top. We should default to results within the last 10 years (&from_date=2012). ( Done) We could also exclude primary evidence here with (category=%281%20OR%2011%20OR%2016%20OR%2018%20OR%2010%20OR%209%20OR%204%20OR%2034%29). (Consensus required whether we really want to hide RCTs. Unsure if there's a way to just hide lower-level primary research.)

• Google Scholar: No way to filter out the low quality stuff but I believe sorting is based in part on citation counts so this should still produce mostly good results. We could add a 10 year filter with (&as_ylo=2012). (This change would affect 'Find general sources'. Could create a separate instance of Google Scholar for 'Find medical sources' but it sounds like we may just want to remove it altogether in favor of PubMed.)

• Set the first or last link to WP:MEDRS pending the addition of some text. ( Done)

• Start working on text that can be included above the link array.

• Possibly revisit links to the paywalled publisher portals (Gale, ScienceDirect, Springer, and Wiley). These were alwasy lower in priority in my mind since existing subscribers already use these services and non-subscribers aren't going to sign up.

• Reconsider GIN. Right now, I'm not getting any useful results. The quotation marks appear to throw off the result set (same applies to JSTOR) but even when guidelines are found, many are inaccessible. The service looks like a work in progress and there may be better guideline collections elsewhere. I need to look.

Just a few ideas. - Wikmoz (talk) 00:31, 14 November 2021 (UTC)

Thank you for all of this, Wikmoz. (This is the kind of analysis/discussion it would have been optimal to have had before this template change went live.) I don't want to chop your very long post, so will add everything here, and hope I don't miss anything. My responses in order of your points above:

If there are areas where we can improve the results by adjusting the search parameters, that is excellent news, but all I am asking for is that we provide some context so that, when a new editor ends up reverted after using a source they found via these links, and asks why, we can say "read the full link, it's not that simple" (MEDRS). If you are able to do more than that, it is gravy.

I agree there is no silver bullet; same applies to RS (as it does to MEDRS). Experienced editors know how to properly use sources. I am only asking for a link so that we have an answer to new or inexperienced editors who are reverted after they think they have followed the advice in this template.

Yes, Mayo is usually good for a sanity check, but we shouldn't leave the impression that is OK to source text to them. For example, on the Morgellons issue, they have taken a bad PR hit from the Morgellons' advocates, and they have clearly adjusted their website to accommodate those advocates ... they are a very inferior source wrt Morgellons, as PR seems to have come in to play. We just need a link to MEDRS, and context, to be able to answer the advocates who use sources from these links.

Re Should we focus on evaluating the utility of source suggestions as they apply broadly to most medical topics or in the context of controversial topics? mixed answer. First, I could go forward with examples from uncontroversial topics if that would help; I am certain I will find similar, so let me know if this would be a wasted effort. And second, it is the controversial topics where this template will do the most damage and sap editing time, as we are then required to give long explanations after reverts. I don't know to what extent same may happen on less controversial topics, but in either case, we solve both by simply providing context.

On your suggestion re each item, I think for all of us to sort all of those out will take some time, and the actual entries may be in flux for some time. IF we provide context, it's OK if we eventually decide some of the links are less useful and we should drop them. (I will be in a better position to opine on that if I go through a sample as above for Tourette syndrome or dementia with Lewy bodies, but don't want to make this unnecessarily long at this point; I am happy to do those samples if it will help.)

For now, we just need some additional context mentioning MEDRS so that we won't see unnecessary confusion to new editors (which can be very offputting), or edit wars in controversial topics where this template now appears to endorse many sources which have been rejected. I have learned not to try to wordsmith additions, as my prose is not stellar, so I agree with you in the hope that more editors will weigh in and we can fine tune some sort of wording linking to MEDRS.

Thanks again, SandyGeorgia (Talk) 01:22, 14 November 2021 (UTC)

Would support excluding primary RCTs in Trip. I don't think they're very useful unless a secondary source addresses them, and then we would just use the secondary source. — Shibbolethink ^{(♔ ♕)} 19:37, 14 November 2021 (UTC)

So at some point recently Google Scholar added a "reviews" parameter but I don't know if it'd be of any use to medical articles; here's a sample for Tourette's. Would that be useful or is it too broad? JoJo Eumerus mobile (main talk) 17:52, 16 November 2021 (UTC)

All of those look excellent, although they would probably all be returned from PubMed as well. (Thanks, Jo-Jo!) SandyGeorgia (Talk) 19:44, 16 November 2021 (UTC)

I admit, part of the reason why I asked about that Google Scholar parameter is because Google does not document it anywhere. Jo-Jo Eumerus (talk) 20:20, 16 November 2021 (UTC)

My conclusion is disappointment. I'm reminded of the battles long ago at WT:MED where editors overloaded the article infobox with external links to random databases they found on the internet. Prime real-estate in our article, taking readers off of Wikipedia and onto sites of variable quality. Here, it seems a bit like search links were added because those might return something useful rather than because all of them are some Wikipedian's favourite way of finding article sources. Only a few of these are imo genuinely useful, the rest are simply a waste of editor time. It seems perhaps a better approach might be to create a project-help page with sections for each potential search engine editors might use, and guidance on how best to use it. Often, the article title is not specific enough to be useful. I'm not at the moment persuaded that this approach of per-article convenience links is a useful technique.

-- Colin°^Talk 11:48, 14 November 2021 (UTC)

Controversial topics

[MOVED FROM BELOW SAMPLES]

I would say it is not particularly fair to base criticism only on the most controversial topics, of which these 3 are extremely controversial. Any good "find sources" box would have issues here. There is no "perfect" solution to source-hunting in difficult and controversial topics, because it requires consensus and human-expert intervention to find good sources on these. I am not sure we will ever find a perfect set of search parameters to make any sources template work for these topics.

It is the very nature and prerogative of pseudoscience to look as much like science as possible. You can't make a Swan-checker that will always find the Goose.

Wouldn't it be better to just hide this on those controversial articles and provide a more targeted search which avoids certain MEDLINE keywords, journals, etc? There's no need to let the perfect be the enemy of the good here. I think it's great that @SandyGeorgia and @Wikmoz are working on ways to improve the serach algorithms. Really, I think that is excellent worthwhile work. But we must be careful to avoid making it worse in some areas while making it better in others. Perhaps the solution is to have a pretty good overall template and then just silence this on areas where the overall landscape of the scientific literature has issues. — Shibbolethink ^{(♔ ♕)} 03:53, 14 November 2021 (UTC)

Shibbolethink when several participants indicated they saw the problem, I stopped generating samples. I will later today add mainstream articles as well, but all that will do is chunk up the page and show the same thing. Stay tuned. My point is that, where this template will cause the most problem for us in here, dealing with the fallout, will be precisely those places where we need to guide editors toward sources. Just please provide a link to MEDRS so we aren’t typing ‘til we’re blue in the face for nought. SandyGeorgia (Talk) 15:16, 14 November 2021 (UTC)

Re: Morgellons: DOAJ and Gale, I actually see quite a few articles (including some systematic reviews) about the nature of this as a delusional parasitosis. Which is the scientific consensus. What's wrong with that?

Overall, this seems to be a case of finding every possible way to hate this change, and not giving it any of the benefit of the doubt. There are many high quality sources in these searches for Morgellons. Ones I would direct my students and colleagues to as well. [2] [3] [4] [5] — Shibbolethink ^{(♔ ♕)} 03:43, 14 November 2021 (UTC)

On review, I think we just need better filters. Excluding primary articles altogether is probably a good idea for DOAJ and Gale, for example — Shibbolethink ^{(♔ ♕)} 19:44, 14 November 2021 (UTC)

What's wrong with that is that the Morgellons advocates explicitly reject that it is actually delusional parasitosis. And, we have a separate article at delusional parasitosis. The very people we need to reach won't read those sources at all (such is the nature of the delusion). SandyGeorgia (Talk) 19:48, 14 November 2021 (UTC)

Then I would say they would never agree with any version of this template. They would only agree with one which supports their worldview, which wikipedia is not meant to do. — Shibbolethink ^{(♔ ♕)} 19:52, 14 November 2021 (UTC)

User:Shibbolethink. You are correct; the editors who give us fits at this article are typically, by definition, working under a delusion. This is why we need the MEDRS link in the "find sources" to avoid going round in circles; I think we have this covered now by adding a link, but this was one of my early samples of why we needed the link-- glad I am no longer having to make that case, and we can now move on to looking at more mainstream articles :) Best, SandyGeorgia (Talk) 20:22, 14 November 2021 (UTC)

Edit Proposals

Remove Google Scholar and Books

Resolved to yes. Google Scholar and Books removed

I'd remove Google Scholar and Google Books from the list entirely. They scrape widely and indifferently. Sorting by recency won't help on controversial topics with recent preprints. XOR'easter (talk) 15:17, 14 November 2021 (UTC) I agree. If we want to have a convenience link for experienced editors, they are great. If we want a list of sources that are MEDRS compliant, they are 99% unsuitable. Jo-Jo Eumerus (talk) 15:25, 14 November 2021 (UTC) Jo-Jo Eumerus, I'm interested in this claim "they are great". Can you give examples of articles where I'd find joy clicking on the majority of these links? And where such joy had to be so immediately received, that having to type the article title into the search box was an issue requiring technology on every medical article talk page. User:Shibbolethink, you also seem supportive. I'm just trying to see if I'm missing something, because at the moment all I can do is feel sorry for those who put so much work into something that isn't even "great" for experts, and is "likely to a problem that causes us a headache" for newbies. -- Colin°Talk 17:03, 14 November 2021 (UTC) Well, I do find almost all my sources on Google Scholar. I know however how to weight sources and exclude bad ones - hence why I qualify with "for experienced editors". To be clear: I don't think we should by default point people to Google Scholar/Google Books, not even on non-medical articles, because not everybody who sees such convenience links knows how to filter them. Jo-Jo Eumerus (talk) 17:17, 14 November 2021 (UTC) Jo-Jo Eumerus, I had a look at your user page and you seem volcanically inclined. Are there medical subjects you've successfully used Google Scholar/Google Books for? -- Colin°Talk 17:50, 14 November 2021 (UTC) Comment. I agree that PubMed is the better solution for searching (and filtering) medical literature. Based on their comments, I think SandyGeorgia and Colin would support removing these two sources as well? - Wikmoz (talk) 17:10, 14 November 2021 (UTC) I’d like to stay out of the weeds in what links are used … mostly because I think almost all of them are useless, and wish we had instead worked on improving what was already at Template:Reliable sources for medical articles. I think we pretty much have a mess here now, and am more focused on fixing the most urgent. SandyGeorgia (Talk) 18:07, 14 November 2021 (UTC)  Done. Google Scholar and Books have been removed per consensus. - Wikmoz (talk) 18:41, 14 November 2021 (UTC)

Sources to include

Rather that having to argue for links to be removed, can we think about it the other way round? The pubmed search can be configured to select reviews/etc and whats more, editors can further configure the results if they know how (or are shown how). Nearly all the relevant/useful articles found by other searches can be found on PubMed (correct me if I'm wrong on that for medical articles). I know there are these things called books, which are also very good, but those usually in my experience involve access to a university library or making a purchase. Has anyone made a convincing case for any of the other links? Even pubmed isn't ideal as it can return so many results that most editors would want to pick other more specific search keywords. For example, on TSC, there are a few dozen terms for the various symptoms and effects that would be good search terms. So the idea that one could improve an article merely by searching sources using the title is limiting. -- Colin°^Talk 17:50, 14 November 2021 (UTC)

We could but I think that discussion will take longer. Just seeing if we can make a few very quick changes that would likely have no objectors and yield an immediate net benefit. - Wikmoz (talk) 18:02, 14 November 2021 (UTC)

I am sympathetic to the notion that some very useful information might be available in some of these links, but there are two kinds of editors: experienced (who don’t need these links anyway) and unexperienced (who will be misled by the whole mess). The only real fix is to eliminate all of it in favor of an improved version of Template:Reliable sources for medical articles. We have a cart before the horse problem here. Meanwhile, I am deeply grateful that the “quick fix” addition of a MEDRS link was added. SandyGeorgia (Talk) 18:11, 14 November 2021 (UTC)

@Colin: to answer your question: I would support keeping the Pubmed and Cochrane links, as these are most likely to be beneficial to new editors. Getting the filters right in Pubmed is a skill I didn't use until my 2nd year of grad school. So it would be great to give new editors those filters/sort-by things on a silver platter. I would also support keeping OpenMD and ScienceDirect, as these typically lead to secondary and tertiary sources that are of high quality. Such as Medline, Statpearls, Medscape, etc.

However, I would also change ScienceDirect to have this query framing instead: Tuberous Sclerosis. Filtering for: Reviews, Encyclopedia chapters, Book chapters, practice guidelines. This avoids much of the cruft.

I also would support keeping DOAJ and Gale, but at the end of the list. I like these databases, but they do require a more expert hand to sort through. They do, however, sometimes turn up things that PubMed ignores or doesn't index. So maybe, due to the "ease of use" and "inexperience - experience" spectrum, I would order it as: OpenMD, ScienceDirect, StatPearls, Cochrane, PubMed, DOAJ, Gale. Just my 2 cents of the ones I have personally found most useful/noob-friendly so far.

The remaining links I do not think are very useful, as they are far too broad and not well-suited to the task of finding sources for medical articles. — Shibbolethink ^{(♔ ♕)} 18:28, 14 November 2021 (UTC)

I have made the edits to the ScienceDirect link I indicated [6] because I'm quite confident no one here would object to them, if that database is kept. — Shibbolethink ^{(♔ ♕)} 19:00, 14 November 2021 (UTC)

To be honest, this discussion is happening at a faster pace and in more places than I can keep up with. Shibbolethink, let me just say I'm deeply sceptical about the claim these links are useful. We need evidence, not just claims. Just as Sandy and I have done, this list of links should not have got off the ground without editors showing for half a dozen varied biomedical articles that the links are all individually useful. It may be that you personally are so academically advanced and have so much time on your hands to browse through pages and pages of junk in order to find one review that for some reason PubMed didn't index. But if that is the case, why on earth do you need someone to type the article title into the search box. Please, those who are proposing these, show us evidence that they are useful. -- Colin°^Talk 17:21, 15 November 2021 (UTC)

Yes to that. This issue sapped my time for three days, and left me plain worn out and discouraged. I am deeply grateful that the people who put this into practice have now engaged, and are taking our concerns seriously, but I really hope that the message received is: next time, please prominently display the exact proposal, with examples, to a wide audience before it goes live across thousands of articles, causing a scramble to fix a problem.

I am relieved that we have made progress such that the template is not as damaging as it initially was, but I'd also be happy if we suppressed "find sources" on all medical articles and instead worked in an upgrade to Template:Reliable sources for medical articles. I recognize that is probably too much to ask of the software, and I acknowledge this fell out of an attempt to fix the November 2020 edit (which was even more misleading), but that edit was also a bad idea. I am still gobsmacked that Wikipedia is sending editors to one and only one news source: The New York Times. SandyGeorgia (Talk) 17:40, 15 November 2021 (UTC)

Link to MEDRS

Resolved to yes. First link is now "Source guidelines," directing readers to MEDRS.

I am abstracting text from above so this won’t be missed: SandyGeorgia (Talk) 17:14, 14 November 2021 (UTC) I know Mathglot has been working on a flexible solution for linking WP:MEDRS in the label. For the moment, I've added it to the source list as the first link. - Wikmoz (talk) 16:30, 14 November 2021 (UTC) Wikmoz, Most helpful! Is it possible to add any plain text along with that, or is that asking too much of the software? If it is possible, we could come up with three or four words that amount to something akin to “read me first”. SandyGeorgia (Talk) 16:54, 14 November 2021 (UTC) Yes. It's possible to add text in parens after a link if there's consensus to do so. Ideally, we'd be able to keep it to two words (editor notice, read first, ideal sources, etc.). - Wikmoz (talk) 17:03, 14 November 2021 (UTC) Yes, I was trying to turn "find medical sources" blue, with WP:MEDRS as the destination, and a nice tooltip, saying something like this. It worked in isolation; that is, we could demonstrate it as is done above using {{medical sources notice}} and it would work, but it breaks when transcluded from the {{Talk header}} template, so I undid it. It can be turned blue without the tooltip, but that didn't seem as useful. There still may be a way to do it, using the "first link" method. Wikmoz, see the sandbox for what's been tried so far. If you can get the tooltip into the "first link", then we'd have it. Mathglot (talk) 20:59, 14 November 2021 (UTC) This goes a long ways towards solving the critical problem. I like “read first”, but “sourcing guideline” or anything similar would also work. Can the MEDRS be bolded? SandyGeorgia (Talk) 17:14, 14 November 2021 (UTC) How about "Referencing standards (read me first)"? XOR'easter (talk) 19:08, 14 November 2021 (UTC) Love it; but I'm not the master wordsmith :) SandyGeorgia (Talk) 19:40, 14 November 2021 (UTC) I like this option, or "Relevant guidelines (read this first)" — Shibbolethink (♔ ♕) 19:40, 14 November 2021 (UTC) That works, too. Now back to my next sample of a high-view article that is not at FA level. SandyGeorgia (Talk) 19:44, 14 November 2021 (UTC) Any way to make this shorter? [ Find medical sources: Source guidelines (read first) · PubMed... ] or [ Find medical sources: Editorial guidelines (notice) · PubMed... ] or [ Find medical sources: Medical source guidelines · PubMed... ] - Wikmoz (talk) 20:11, 14 November 2021 (UTC) If those three are the only choices, the way to shorten is to use the third and ditch the first use of the word medical, as it is used in the link. But Colin and WAID are the wordsmiths here, so what do I know. SandyGeorgia (Talk) 20:27, 14 November 2021 (UTC) Those aren't the only choices. Just throwing out some shorter options. Yes, if we consider the label text, something like this could work: [ Find sources: Medical source guidelines · PubMed... ] or [ Find medical sources: Source guidelines · PubMed... ] - Wikmoz (talk) 21:10, 14 November 2021 (UTC) Either works, I am indifferent, thanks! SandyGeorgia (Talk) 21:13, 14 November 2021 (UTC) Honestly just "guidelines" is enough for me — Shibbolethink (♔ ♕) 21:20, 14 November 2021 (UTC) That could be confusing since it could be read as a link to related treatment guidelines. - Wikmoz (talk) 21:31, 14 November 2021 (UTC) Mathglot, permissions were escalated since this is a high use template. I've requested access but in the interim, can you change this to the following. Probably won't be the final text but will help to see this in place. { code = 'medrs', display = 'Source guidelines', tooltip = 'Identifying reliable medical sources *READ FIRST*', separator = ' (notice) · '}, - Wikmoz (talk) 21:41, 15 November 2021 (UTC) @Wikmoz:, I put it in the sandbox, so you can play with it and/or discuss it here, until you/everybody is happy with it, and then can move it to the module config (or I can at that point). Live on top, substed sandbox rev. 1055462546 on the bottom. (I added the bold since someone had mentioned that somewhere.) Find medical sources: Source guidelines · PubMed · Cochrane · DOAJ · Gale · OpenMD · ScienceDirect · Springer · Trip · Wiley · TWL Source guidelines (notice) · PubMed · Cochrane · DOAJ · Gale · JSTOR · OpenMD · ScienceDirect · Springer · Trip · Wiley · GIN · WP Library Thanks, Mathglot (talk) 00:57, 16 November 2021 (UTC) Nice work on the bolding hack! Subst is definitely helpful here. I think if we bold the link then that visually differentiates it from the others so "(notice)" would no longer be needed. Mockups (links don't function): Find sources: Medical source guidelines · PubMed · Cochrane · DOAJ · Gale · JSTOR · OpenMD · ScienceDirect · Springer · Wiley · WP Library Find medical sources: Source guidelines · PubMed · Cochrane · DOAJ · Gale · JSTOR · OpenMD · ScienceDirect · Springer · Wiley · WP Library - Wikmoz (talk) 01:12, 16 November 2021 (UTC) [Link sequence fixed with thanks to Mathglot] @Shibbolethink, XOR'easter, SandyGeorgia, WhatamIdoing, Talpedia, and JenOttawa: MEDRS is now the first link in the template. We're planning to rename the link. Several options appear above along with recent mockups. If there's some consensus that one of these is a clear improvement over "MEDRS" then we can move forward immediately. We can always workshop something better in the future. - Wikmoz (talk) 02:02, 16 November 2021 (UTC) They both look fine, I am still indifferent, but thrilled you all have made the fix (and recognizing it was lots of work). SandyGeorgia (Talk) 02:07, 16 November 2021 (UTC) (edit conflict) Wikmoz, there's a discussion somewhere above where you were talking about free vs. paywalled sites, and that it's more complicated than just a binary thing. So, I thought of maybe using the padlock icons to indicate varying gradations of accessibility. I assumed we couldn't put those icons into the config, but it turns out we can. Here's sandbox rev. 1055472044 with a padlock and custom tooltip on Springer, based on my recollection of what you said (might be off): Find sources: Medical source guidelines · PubMed · Cochrane · DOAJ · Gale · JSTOR · OpenMD · ScienceDirect · Springer  · Trip · Wiley · WP Library Maybe we could use icons to provide users a heads-up about what to expect regarding accessibility at the different links? Sandy if you're lurking, there's lots more of these icons; see here for example. Mathglot (talk) 02:15, 16 November 2021 (UTC) updated tooltip to rev. 1055477535; Mathglot (talk) 02:56, 16 November 2021 (UTC) Yes, lurking, because ... dealing with student edits. I despise those icon-y things :) They clutter up text in article citations, mean little to the uninitiated, and most experienced editors know their way around most of them anyway. I also hate how often my content work is disrupted by bots that go around updating the darn things. Probably not what you wanted to hear :) I'm happy with whatever y'all come up with, as long as we have MEDRS in there. I'm sure someone likes those icons, or we wouldn't have 'em, so ignore me! Best, SandyGeorgia (Talk) 02:20, 16 November 2021 (UTC) I don't have a dog in the race, so I'm happy hearing anything that gets us closer to the consensus of what medical editors want, and will aid new and experienced users in finding the right sources. Mathglot (talk) 02:24, 16 November 2021 (UTC) Glad (and kind of surprised) to see that we have so much flexibility. Regarding the lock icons, given how muddy the distinction is and how much space a bunch of these icons would take up, I'd favor keeping those reserved for documentaion, where we can also fit a key. Really good to know the option exists though. I could imagine other use cases for using icons to differentiate a special link (e.g. a list icon next to the MEDRS link) in the future. - Wikmoz (talk) 02:54, 16 November 2021 (UTC) I concur. Just good to know it's available, if we need it. Mathglot (talk) 03:00, 16 November 2021 (UTC) For the record: now I'm not quite sure it is available; see this edit. It may be fixable, but per comments above it's not necessary to look into that now. Mathglot (talk) 18:48, 16 November 2021 (UTC) Moved your code into the separator field and it appears to work. - Wikmoz (talk) 20:42, 16 November 2021 (UTC) Mathglot, can you sync the current sandbox (minus the padlock) to production? - Wikmoz (talk) 20:48, 16 November 2021 (UTC) @Wikmoz: I’m mobile until Thursday late, so I actually can’t right now. Maybe Sdkb can? I can respond at Talk, but I’m shy about doing any template or module changes for fear of screwing it up while thumb-typing. Mathglot (talk) 02:54, 17 November 2021 (UTC) Or just request permissions and link this TP section. They sometimes act pretty quickly on such requests. Mathglot (talk) 03:01, 17 November 2021 (UTC) Sorry, I haven't been following this closely. What is production and which sandbox is this? {{u|Sdkb}} talk 03:39, 17 November 2021 (UTC) Just hoping to copy the contents of Module:Find sources/templates/Find sources medical/sandbox over to Module:Find sources/templates/Find sources medical. I'll be online for another hour so if you are able to do this, I can test everything on live immediately after the move. - Wikmoz (talk) 03:50, 17 November 2021 (UTC) No worries. I did request permissions yesteray. The request just hasn't been answered yet. - Wikmoz (talk) 03:27, 17 November 2021 (UTC) An admin changed the permission level to EC so I was able to push the change. User:SandyGeorgia, I hope this helps. If there are no immediate change requests, I'll collapse this thread and we can shift the focus to other items. - Wikmoz (talk) 20:36, 18 November 2021 (UTC) @Wikmoz:, glad you were able to resolve this one. Back now if you need support again. Also, I like the collapsing; feel free to use {{done}} or any of the other symbols here that are useful. Mathglot (talk) 04:05, 19 November 2021 (UTC) Happy dance, thanks to all ! SandyGeorgia (Talk) 04:17, 19 November 2021 (UTC)  Done Marking this one as closed and collapsing. Mathglot... that page is great! Going to save me a lot of time in the future. Thank you - Wikmoz (talk) 05:17, 19 November 2021 (UTC)

Move PubMed to first link in list

Resolved to yes. PubMed link repositioned.

Colin noted above that he was surprised to see Cochrane as the first link displayed in the template. No objection from me to moving it to the front (after MEDRS). If there's consensus to do this, it's a very easy change. - Wikmoz (talk) 17:48, 14 November 2021 (UTC) Per my note above, I'm needing some convincing evidence that editors writing biomedical content actually find any of the other links to be commonly useful, with nothing more than the article title in the search box. The rest, as far as the evidence presented so far, all seems to steer the editor into the random wilds. -- Colin°Talk 17:53, 14 November 2021 (UTC) Understood. I'd agree with your assessment if we were directing medical topic editors to Google and Facebook, which are the real alternatives for many editors. Indeed, that's in part what we were doing for a year before this update. Obviously, there's no way to automatically create a detailed query but I do think it's helpful to provide editors with tools/options that point them in the right direction and let them drill down or modify the search query as they see fit. I think we lose the forest for the trees a little by focusing on the worst case scenario for contested topics, where the template can be easily hidden. - Wikmoz (talk) 18:46, 14 November 2021 (UTC) The link, even if not ideal, may also be a time-saver. Even if one knows that the article title isn't the ideal query, after clicking it and getting the not-very-good results, you're now on the search results page with the search box at the top and prepopulated with the query, so you're in the right place to refine it, or retype the query entirely. Beats looking around for my b:rowser bookmarks and having to type everything from scratch. I.e., it caters to my laziness and forgetfulness: which to me is valuable, but I realize others may have different work methodologies. Mathglot (talk) 20:44, 14 November 2021 (UTC) Yes this is the other reason I liked this. I use these databases all the time, having their links on the talk page is just straight up useful. — Shibbolethink (♔ ♕) 21:21, 14 November 2021 (UTC) I have occasionally found something in Google Books that I couldn’t find elsewhere, but see my comment above re two kinds of editors: an experienced editor may need to and know how to use Google Books, while the inexperienced editors will be seriously misled by the sources they find in this template. I do agree that moving PubMed to the first slot might mitigate the immediate effects of this template, while we work out a better solution; right now, we’ve got potential for damage throughout medical talk pages, and immediately getting PUbMed first might solve some of the short-term problem, but not solve the big-picture problem. SandyGeorgia (Talk) 18:14, 14 November 2021 (UTC) As to the bigger picture, I will next work up more mainstream samples, so we have a larger sample upon which to base these decisions. Back in a few hours, SandyGeorgia (Talk) 18:18, 14 November 2021 (UTC) Also, Colin I am unsure if you saw that, one reason we ended up here is that in November 2020, generic links were added to all talk headers, which was even more misleading for medical content. I think this happened because of attempts to fix the first mistake, which resulted in an equally bad second mistake. I am wondering how useful the initial November 2020 generic links are, and whether other content areas are having the same problem, but have not had time to research that. Pick any non-medical topic you know well that has a talk header, and have a look if you are interested in following that up. There remains a possibility that the whole thing, back to Nov 2020, might need attention. SandyGeorgia (Talk) 18:24, 14 November 2021 (UTC) EG, I just looked at Talk:Nicolás Maduro and had a holy cow moment! Wikipedia is giving some sort of preferential treatment to the New York Times. What the holy heck, and where is the consensus upon which that was based? SandyGeorgia (Talk) 18:27, 14 November 2021 (UTC) @Sdkb: I think this is your cue, re newspaper of record. Mathglot (talk) 21:16, 14 November 2021 (UTC) Yep, I'll reply here. {{u|Sdkb}} talk 21:22, 14 November 2021 (UTC) And now moved to here. Regards, SandyGeorgia (Talk) 21:41, 14 November 2021 (UTC)  Done. Without objection. PubMed moved to first position after MEDRS. It's trivial to revert if anyone objects later. - Wikmoz (talk) 18:56, 14 November 2021 (UTC) Great, SandyGeorgia (Talk) 20:28, 14 November 2021 (UTC)

Add StatPearls link

Personally I would support adding a "Statpearls" link, as these articles are of high quality imo. Such as these queries: Tuberous Sclerosis  · Chronic Lyme  · PANDAS. It's a bit like a cheaper (i.e. free) but still useful version of UpToDate without as much guideline/treatment info. Great for noobs to find good sources also, as StatPearls' contributors tend to cite many high quality publications.

Think of it like a wikipedia written by doctors for other doctors. Except each article is peer reviewed and published as an ongoing updated thing by its authors, who are selected as experts on that topic by a panel of their peers. When I encounter a term I haven't heard before in the course of treating patients, StatPearls is the first place I punch it in. It's high enough quality that doctors in America get continuing education credits for reading it. It's also published under the creative commons license!

I have created the appropriate linktable, doc, etc. for this StatPearls query: [7] [8] and put it on the list of linktables: [9], but I have no idea how to add it to this template, if we gain consensus to do so. On that, I would defer to @Wikmoz. — Shibbolethink ^{(♔ ♕)} 19:11, 14 November 2021 (UTC)

Nice work on the module edits!! If this gains consensus, adding to the medical link list is easily done here. Is the NCBI bookshelf view preferred? It looks like the content is also available at StatPearls.com. Presentation is about the same but the search results are ranked a little better.

• https://www.ncbi.nlm.nih.gov/books/NBK430685/?term=covid
• https://www.statpearls.com/search/sitesearch?q=covid

Wikmoz (talk) 19:27, 14 November 2021 (UTC)

Thanks :) and thanks for the explanation re: module. I couldn't find that page, but I also didn't look very hard! I'll happily add it if it gets consensus

It's definitely a dealer's choice sort of thing re: NCBI. I tend to like the NCBI interface because it doesn't include any of the "subscribe to me" cruft that sometimes appears on the official site. But the official site also includes the multiple choice questions which are often pretty good. And a better presentation of the authors and their COIs, if any.

I would support either one — Shibbolethink ^{(♔ ♕)} 19:30, 14 November 2021 (UTC)

could we hold on Stat Pearls until one of us has time to dig back into archives, about a year or so ago, when we had a big discussion of StatPearls? SandyGeorgia (Talk) 19:52, 14 November 2021 (UTC)

You mean these? [10] [11] [12] [13] I would agree with the general consensus in these that StatPearls is useful for a broad overview, and they sometimes link to older articles. But I would say they rarely if ever link to bad sources or provide explicitly incorrect information. Sometimes just briefly outdated (e.g. the Chronic Pain article is now much better than suggested by one of those past discussion participants). And, for these reasons, I think it is a good place to send newbies. We use it in my hospital all the time.— Shibbolethink ^{(♔ ♕)} 19:59, 14 November 2021 (UTC)

No, there was a different discussion; let me see if I can find it. (I can't even remember what we concluded, btw.) SandyGeorgia (Talk) 20:30, 14 November 2021 (UTC)

Perhaps it was the one started by PainProf Sorry no that's above CV9933 (talk) 20:45, 14 November 2021 (UTC)

OK, I can't find it, so ignore me. My memory is of finding a StatPearls entry that was misleading, dated or inaccurate ... but we are going to find that anywhere we look anyway (no source or database is perfect), so I have no basis to object. SandyGeorgia (Talk) 20:56, 14 November 2021 (UTC)

I have previously expressed concern about StatPearls based on their coverage of psych topics, but I must say that over the past year or so, I have read only very well-written, accurate StatPearls articles. Hopefully that means they have good quality control mechanisms in place. :0). Mark D Worthen PsyD (talk) [he/his/him] 19:09, 17 November 2021 (UTC)

@Shibbolethink:, can you clarify if you are talking about searching the StatPearls book at the NCBI bookshelf (bottom link below; gray button marked 'search this book' on this page), or the NCBI bookshelf (top link below, and blue button marked 'search' at that page)?

• NCBI bookshelf search for "myocardial infarction"
• StatPearls book search for "myocardial infarction"

or something else? Thanks. Mathglot (talk) 21:39, 14 November 2021 (UTC)

@Mathglot: StatPearls book. But the StatPearls site itself is basically the exact same content in another form. — Shibbolethink ^{(♔ ♕)} 21:41, 14 November 2021 (UTC)

Thanks! Mathglot (talk) 21:42, 14 November 2021 (UTC)

I've added the link to the sandbox. It sounds like there's general support for this.

- Wikmoz (talk) 05:38, 19 November 2021 (UTC)

Thanks! :) — Shibbolethink ^{(♔ ♕)} 23:04, 23 November 2021 (UTC)

Treatment guidelines link

If there are no objections, I'd like to remove GIN at least for now because the quotation marks that are automatically added by the template break the search. If/when that can be fixed, we also need to add the English language filter. Alternatively, is there a better option? NICE Evidence has a good search. Not the broad reach of GIN but it's a known quantity. - Wikmoz (talk) 21:54, 14 November 2021 (UTC)

No objection from me. SandyGeorgia (Talk) 22:01, 14 November 2021 (UTC)

JenOttawa, any comment? Mathglot (talk) 22:11, 14 November 2021 (UTC)

Wikmoz, tech note: not sure if those module config pages can have embedded templates (I've never tried) but if they can, we could remove the quotes using {{replace}}. Alternatively, {{find medical sources}} could have a filtration switch for known sites like this one (or even a set of config subtemplates as at the module) and move the query terms from positional arg1 to arg2 (unquoted search) for GIN, which should solve the problem. (You still have to have an arg1, I believe, but you could probably pass a blank.) Or, just "replace" arg1. Mathglot (talk) 22:17, 14 November 2021 (UTC)

Worth a shot! It looks like the template access was just restricted so I can't edit but I guess https://guidelines.ebmportal.com/guidelines-international-network?fv%5B65%5D%5B0%5D=1579&g=collection_field_3&type=search&search={{replace|$1|"|}} would be the approach... added the English-only filter in there as well. I'll request additional permissions if I get blocked from other modules. - Wikmoz (talk) 22:34, 14 November 2021 (UTC)

Absolutely no objection to remove. This was just an idea, not 100% certain if it is suitable. Thanks for checking in Mathglot (talk · contribs). JenOttawa (talk) 02:31, 15 November 2021 (UTC)

Mathglot, I can't edit the GIN module but tried it with another link and Lua returns an error. Going to remove for now since the current search doesn't work and there doesn't seem to be a quick fix. - Wikmoz (talk) 04:27, 15 November 2021 (UTC)

I would support NICE over GIN. much more well respected. — Shibbolethink ^{(♔ ♕)} 01:25, 16 November 2021 (UTC)

UpToDate link

Sandy, above you said that UptoDate requires a subscription, but I'm getting results, for example for "Tuberous sclerosis". Does a subscription get you wider results, and free search is more limited? Mathglot (talk) 22:09, 14 November 2021 (UTC)

Most of the results are paywalled (after the first few paragraphs). The exception might be their patient education content. Would be amazing if Wikipedia Library could provide access to this or Epocrates. - Wikmoz (talk) 22:14, 14 November 2021 (UTC)

Oh I see. There's a "request a database partner" link somewhere at Wikipedia Library; definitely worth a try. Mathglot (talk) 22:19, 14 November 2021 (UTC)

Yep ... they give you just enough to entice you to pay the heavy fee for more :) SandyGeorgia (Talk) 22:51, 14 November 2021 (UTC)

Paywalled links

In the spirit of Colin's "what to include" (instead of what to remove), I thought I'd throw this out there for consideration: instead of removing everything that is paywalled, what about keeping them if they're good, but just demoting them instead? I could see fronting all the completely free links, and then maybe move those that require subscription or have other restrictions to the back, possibly in smaller, grayish font, possibly in parentheses, with some kind of caveat label. This leaves newbies and unsubscribed senior editors with all the accessible links first, but still provides links to good resources unobtrusively at the back for those that have access, without having to hunt down their bookmarks file. Just a thought. Mathglot (talk) 22:40, 14 November 2021 (UTC)

Except a lot of what they will find at PubMed will be paywalled, and it is still good to have it first. (I may be misunderstanding what you are proposing.) SandyGeorgia (Talk) 22:52, 14 November 2021 (UTC)

The services fall along a spectrum. Some of the links (DOAJ) return 100% free results while others (ScienceDirect after lockout) will return 0%. PubMed and Trip are somewhere in the middle (though we could add a "(free)" filter link next to PubMed. The publisher portals (Gale, Elsevier ScienceDirect, Springer, and Wiley) seem to skew more heavily to the paid results. Springer may be the worst. I generally favor the services that provide the most freely-accessible content.

The split that Mathglot is proposing would definitely make it less likely that we direct unsuspecting editors face first into a paywall but the segmentation isn't crystal clear (unless we group them within a "Publisher" bracket). If we are reevaluating which links make the cut, I'm less confident in the benefit of the publisher portals as those with institutional access for whom the links would be most useful presumably know about the services, and the free content will already be indexed in PubMed. - Wikmoz (talk) 00:08, 15 November 2021 (UTC)

I don't know if this is the right place to mention it, but another thing to consider is the Wikipedia library. I'm aware of some of the concerns about it, but there is more than one way to access it. There's the main link, that is, the one that returns results from an assortment of the dozens of collections they provide according to who-knows-what relevance sorting algorithm to provide their top ten result set, but WP Lib also provides individual links to each partner, and I'm pretty sure you can query each one directly. So that if you go through the Wikipedia Library JSTOR link, say, you get only the top ten JSTOR results and not the mix you get from the main link. The point being, since WP Lib is offered to all (registered) users for free, this bypasses the individual registration requirements required by the publicly facing url. I have not tested this, to see if there are restrictions of the same or a different kind than the public site, but I wanted to throw this out there, as a possible alternative approach to a couple of the restricted collections we are talking about, that are also offered by WP Library. The list of partners is here]. Mathglot (talk) 02:44, 15 November 2021 (UTC)

My main concern has always been that we get a MEDRS link added and explained; sorry to have run out of steam on this, but I just had to take a detour to deal with a sockmaster at Tourette syndrome, and I’ve run out of steam here after three days of work. Please ping me if more feedback from me is needed, as I have hit a wall of exhaustion; thanks so much for getting us now to a better place. SandyGeorgia (Talk) 02:48, 15 November 2021 (UTC)

I didn't realize that the resource list was that extensive. That's kind of impressive. I had the "Library Bundle" access (EBSCO and JSTORE are included) but haven't ventured to see what's available beyond that in a while. It looks like you have to individually apply to access some of the more expensive resources and approval takes several days but medical editors would benefit from knowing that these resources are available. - Wikmoz (talk) 04:14, 15 November 2021 (UTC)

That's right, there's a basic bundle of two or three dozen that you get automatically, and the rest you have to apply for. When I started, you had to apply for every one individually, and nothing came automatically, so this is a much better system now, which gives everyone a starting point from which they can apply for others, if they wish to. Mathglot (talk) 06:36, 15 November 2021 (UTC)

The Wikipedia Library is offered only to experienced editors (500+ edits plus some other requirements). Since I expect these links to be primarily used by experienced editors, then I think that this is a good idea. If you wanted to get fancy, I believe that it's possible to display one bit of text to newcomers and a different thing to extended-confirmed editors. This wouldn't be a perfect match for TWL's requirements, but it would work in most cases. WhatamIdoing (talk) 16:54, 15 November 2021 (UTC)

Good point, and interesting approach. For the record, WaId is talking about Template:If extended confirmed. Mathglot (talk) 01:35, 16 November 2021 (UTC)

Remove JSTOR (pending quote fix)

I was reviewing Sandy's findings and it looks like the quotation marks throw off JSTOR more than I realized. I think this is another one that we can safely remove pending a future fix. - Wikmoz (talk) 08:42, 16 November 2021 (UTC)

Since param2 is unquoted search, we may be able to fix this by simply moving the default title or param1, if given, into param2, and blanking param1. Mathglot (talk) 18:54, 16 November 2021 (UTC)

Have shelved that for now; see User talk:Mr. Stradivarius. Mathglot (talk) 02:11, 17 November 2021 (UTC)

Mathglot, thank you for investigating! I've removed JSTOR per your note and removed the padlock, which appears to work as a separator. Can you sync the sandbox to production? - Wikmoz (talk) 02:52, 17 November 2021 (UTC)

See here. Mathglot (talk) 06:00, 17 November 2021 (UTC)

 Done. Removed from list. - Wikmoz (talk) 05:21, 19 November 2021 (UTC)

Followup – why doesn't it appear on this page?

Mathglot why is it not turning on at Talk:Multiple sclerosis? It is showing up on the other med talk pages I checked … SandyGeorgia (Talk) 04:26, 19 November 2021 (UTC)

@SandyGeorgia:, that's because there are a variety of aliases for the WikiProject Medicine" (fifteen of them, actually) and the first version checks only for the original name. M.S. doesn't use that name, but the alias {{WPMED}} instead, and the live code doesn't detect it (yet). This is a known issue, and the sandbox already contains code for some aliases, although not this one yet, and will be added and released in due course, once the most important issues with the first version are resolved. Mathglot (talk) 04:37, 19 November 2021 (UTC)

In the short term, the manual workaround would be to rename {{WPMED}} to {{WikiProject Medicine}} on the Talk page. - Wikmoz (talk) 05:32, 19 November 2021 (UTC)

OK, thanks, will now what to watch for now, and how to fix it. SandyGeorgia (Talk) 05:57, 19 November 2021 (UTC)

That works one-by-one, but there are another 15,000+ that use WPMED, so we'll have to address it in the wrapper. There are other redirects, not all of which require fixing; the complete list is here if you're curious. Mathglot (talk) 06:54, 19 November 2021 (UTC)

Wow, what a ridiculous timesink. Some clever person who knows how to deal with redirect discussions should initiate a move to get rid of the lesser used ones. SandyGeorgia (Talk) 10:50, 19 November 2021 (UTC)

Redirects are cheap, and rarely worth getting rid of. The practical solution, is to handle the popular ones, like WPMED, and to effectively get rid of the unpopular ones by removing all their uses: i.e., by editing the articles that employ them and change them to use the real name, or one of the popular redirects like {{WPMED}} instead. That easy and quick; it's the redirect discussion itself that is the actual time sink. Don't worry, this will get resolved fairly soon. Mathglot (talk) 02:19, 21 November 2021 (UTC)

Thanks again,, SandyGeorgia (Talk) 02:22, 21 November 2021 (UTC)

Np. Just for the record, technical aspects of this issue were first described here, and test cases are here. Mathglot (talk) 02:54, 21 November 2021 (UTC)

Code to handle this has been added to the sandbox, and is now in testing. Mathglot (talk) 03:24, 22 November 2021 (UTC)

Are there things I should know before commissioning medical illustrations?

Hello! I have been interested for a while in improving the images on wikipedia's articles about transgender-related surgeries. A while ago, for example, I guided an acquaintance through the process of creating and uploading photos of their penile-inversion vaginoplasty. I have had a financial windfall and would like to commission a medical illustrator to draw some surgical diagrams and release them with a Creative Commons CC-BY-SA license, to better inform readers about how these surgeries are carried out. I have found some good candidates for professional artists and plan to start with metoidioplasty. Are there any special policies I should be aware of and alert my artist to? I don't see any information in WP:MEDRS which appears to apply, but this is not an area of wiki expertise for me. Thank you! ~ L 🌸 (talk) 21:17, 23 November 2021 (UTC)

@LEvalyn, you might find these pages useful background:

• Wikipedia:Manual of Style/Images
• Wikipedia:Offensive material
• Wikipedia:WikiProject Sexology and sexuality/WIP-image-guidelines (marked historical, but still contains some useful information)
• Wikipedia:Graphics Lab or the equivalent group at Commons, c:Commons:Graphic Lab.

We don't have any guidelines that provide useful/normal information that a graphic designer might expect, such preferred file formats. If there are text labels in the drawings, it's a good idea to use SVG file formats (because they can be translated! See c:Help:SVG#Translations) and also to produce two versions (one with words and one with numbers, so that a low-tech translation can be provided in the captions, by writing "(1) Name of whatever is at 1").

You might want to specify race/skin color, especially if you don't want them all to be the same. WhatamIdoing (talk) 23:54, 24 November 2021 (UTC)

Thank you, WhatamIdoing! Those are useful. ~ L 🌸 (talk) 04:03, 25 November 2021 (UTC)