Should entry/tag names use "&" or "and"? E.g., should it be "sentience & consciousness" or "sentience and consciousness"?
It seems like recent entries/tags - since the announcement of the wiki, not just tags - are mostly using "and". And maybe that looks more encyclopaedic. So I guess we should go with that?
There are many entries/tags for which there is a very related EA-aligned Facebook group (see here for a directory to such groups). I think that, in all such cases, it'd probably be good to add a link to the relevant group in an External links section in the entry/tag. (This is for the obvious reason that many people checking out the entry/tag might also find the Facebook group useful, e.g. to find more info, get career advice, or share relevant drafts for feedback.)
Before I go adding all those links, does anyone think that that's a bad idea?
As for how related the Facebook group should be in order for it to be linked to, the rule of thumb I'm leaning towards is "Each Facebook group should only be linked to from one entry/tag. If the Facebook group seems relevant to multiple entries/tags, that suggests that those entries/tags should probably link to each other in the Related entries section, and the Facebook group should just be linked to from the most relevant one. This should still allow the readers most likely to be interested in the group to find it."
E.g., I'd link to the Risks of astronomical suffering (s-risks) Facebook group from the s-risks entry; but I wouldn't link to it from the entries on suffering and pain, existential risks, trajectory changes, the long-term future, mind crime, etc.; but I'd ensure many of the latter entries link to the s-risks entry.
(I also plan to post in the relevant Facebook groups to mention that the relevant tags/entries exist, so people can check out the posts, tag more posts, or edit the entries [though I'll skip groups where I did this already last year or this year]. But I feel more confident that that makes sense to do, and it's not really a matter for this style guide anyway since it involves promoting rather than editing the entries.)
I don't a strong opinion either way. Part of me wants to encourage people to use the Forum to conduct these discussions, rather than point them to a group on a closed social media platform with mediocre editing and commenting features, especially insofar as this to a significant extent happens because of network effects and trivial inconveniences (rather than revealing an inherent preference for one venue over the other). I also worry slightly that linking to Facebook may make the entries look a bit amateurish and not intellectually serious. On the other hand, I recognize that such links could help users find relevant content and connect them to others with similar interests. I'd like to know what others think before we make a general decision on this.
(By the way, I'm using 'entry' as a neutral term to refer to what may, depending on what functionality is emphasized, be either called a 'tag' or a 'wiki article'. We may want to use that term in the future.)
What should our norms be, or what is the legal situation with, copying text from LessWrong wiki entries verbatim as the text for a Forum wiki entry?
I guess we should probably paraphrase and cite them, rather than copying verbatim, but maybe we don't have to and sometimes it's better (to save time) to at least allow a verbatim copy?
Should there be a norm where, for (almost?) every Forum wiki entry that has a corresponding Wikipedia article, the Further reading / Bibliography / External links section includes a link to that Wikipedia article?
I'd be opposed to such a norm. Very often, Wikipedia is not the best reference on a given topic, and their articles are already extremely easy to find. I would decide whether to cite them on a case by case basis, with a relatively high bar for citing them.
(A rule of this sort may be more plausible with reference works of exceptional quality, such as the Stanford Encyclopedia of Philosophy, but even then it doesn't feel to me like we should have such a rule. I guess the underlying intuition is that I don't see a reason to deviate from the general principle that decisions on whether to include a particular work in the bibliography should be based on an assessment of that specific work's quality and relevance, rather than on some general rule.)
Very often, Wikipedia is not the best reference on a given topic, and their articles are already extremely easy to find.
These are good points I hadn't really considered.
I think what I had in mind was something like this: In cases where the topic covered by a Forum wiki entry is also covered by a Wikipedia article or an SEP article or whatever:
It seems reasonable to ask "Wait, why should there be an EA Forum wiki entry about the exact same topic?"
The main reason this is a reasonable question is primarily that those other encylopedias are fuelled by much more effort than ours is.
(That said, I think it does still often make sense to have a Forum wiki entry on a topic, either because that's also a tag, or to highlight that this topic is relevant to EA stuff, or because EAs might be interested in a specific aspect or implications of the topic that other sources don't cover or don't emphasise.)
It also seems like the existence of those articles in other encylopedias pushes against investing as much effort into expanding and polishing the corresponding Forum wiki entries as we should invest in entries that don't have an existing entry in another encyclopedia.
E.g., I think it would be weird for us to write a huge article covering all the just generally useful info on "Policy change" that one could find on Wikipedia. Mainly because that's just like a waste of effort.
It seems like linking to a larger entry on the topic from another encyclopedia might (a) signal that we're aware that that exists, and we're not just weirdly and ignorantly reinventing the wheel but rather presumably have some reason for making our own thing too, and (b) allow people to find that more detailed info on things we won't write loads of details about.
(I wrote that quickly; not sure I explained it well.)
But, as mentioned, I think your points are good too, so now I'm unsure what norm I'd favour here.
I very much agree with your points, especially the first one. Perhaps as we gain more experience, we can get a better sense of the types of articles that warrant a policy of linking to some external source by default. I can imagine that being the case e.g. for core philosophy topics (e.g. 'normative ethics') and the Stanford Encyclopedia of Philosophy.
And what should the norm be for cases where there's a corresponding LessWrong wiki entry / concept / tag? So far, I've been including a link to that in the Further reading section in (almost?) any case where the Forum wiki entry has little text and/or the LessWrong entry has a bunch of text. But it's plausible I should do this more or less than I currently do.
The principle I've been following is to treat LW like I do all other sources, and cite their articles iff they seem worthy of inclusion. (I do think it's always worth checking out the LW tags, because they are more likely to pass that test than most other sources.)
Is this style guide the right place for policies/norms about how to use tags? E.g., a policy about which posts should be tagged with a tag for an organisation, as discussed here?
Or is there/should there be some other place for such policies/norms?
Seems like that's more about "tagging" and less about "style for the wiki entries".
This doesn't seem like the right place. I've written my thoughts on tagging norms in a few places, but it would be good to collect that (plus others' thoughts) in one place. I'll talk to Pablo and see about one of us creating this resource.
Meanwhile, if you have thoughts on tagging policy, feel free to mention them in replies to this comment (or to create a question post to collect others' thoughts — though I'm not sure how much discussion is required here, as I think our tagging norms/policy will end up being pretty simple).
I think the Style Guide is not the right place for discussion of these norms. As you note, this concerns tagging rather than Wiki content. I'd tentatively suggest commenting on the proposed tags thread, but I'll ask Aaron.
External links should only be used in the Bibliography and External links sections of the article, and never in the lead or body sections (see the Organization of articles section).
I think external links should be allowed in the lead or body sections. And in discussions on a draft of this style guide, I seem to recall that you (Pablo) moved to agreeing with that view? (ETA: As discussed in replies, it turns out I misremembered that.) Should that line of the style guide be changed?
(Let me know if I should explain my reasoning for this position again.)
I don't recall agreeing to this. As I wrote in a comment to the Google Doc:
I am personally quite reluctant to have external links in the body of an entry. I think this looks unencyclopedic. Maybe this is colored by my experience editing Wikipedia, but I'd like to explore the technical approaches described above before considering a relaxation of the "no external links" rule.
The approaches alluded there are to allow users to insert citations entries from a database rather than having to enter the bibliographic details manually. This should make the insertion of a citation roughly as easy as the insertion of a link, so requiring one over the other shouldn't feel constraining. I understand that the tech team is very busy with other stuff, so I'm not sure whether we can expect any of these innovations to be developed soon (or at all), but for the time being this work can be delegated to our assistant, and we will make it clear in our FAQ users should not waste any time formatting citations. If the situation doesn't change in 3-6 months, we can reconsider.
I realize I probably haven't made this very clear, but a core principle I've been following when writing content and planning (and one that I feel reluctant to abandon, given its centrality to how this project was conceived) is to follow existing scholarly norms as closely as reasonably possible. I believe this is generally desirable insofar as many of these norms can be observed at little or no cost, while providing significant benefits in the form of recognition and authoritativeness, but is especially important for a Wiki that will cover areas that are in many ways speculative and are either neglected or not held in especially high esteem by academia, journalism, and other elite institutions.
So my opposition to having external links in the body of an entry is based on my perception that this is generally not an accepted scholarly practice. Wikipedia doesn't normally allow the use of such links except in special sections (such as "external links" and references sections), and my casual acquaintance with other encyclopedias and reference works suggests that they too follow norms of this sort. If you or others could point me to counterexamples sufficient to establish that this is not in fact common, I would be happy to revise the policy, to which I have no personal attachment.
Ah, yes, I just checked the doc comments and you're right. I must've simply misremembered.
I realize I probably haven't made this very clear, but a core principle I've been following when writing content and strategizing (and one that I'm reluctant to abandon, given its centrality to how this project was conceived) is to follow existing scholarly norms as closely as possible. [...this] is especially important for a Wiki that will cover areas that are in many ways speculative and are either neglected or not held in especially high esteem by academia, journalism, and other elite institutions.
That makes sense, and I think I'd underestimated that argument in favour of not allowing external links. So I'm now less sure what we should do here.
If you or others could point me to counterexamples sufficient to establish that this is not in fact common, I would be happy to revise the policy, to which I have no personal attachment.
I don't have any counterexamples. Instead, I'll point to some reasons why I think allowing external links may make more sense for our wiki than for Wikipedia:
We're way, way smaller.
Apparently English Wikipedia has over 6 million articles, whereas it looks like our Tags Portal lists ~500 tags.
If Wikipedia limits itself to only internal links or references in the body of an article, that still allows a link to an article or section on a substantial portion of all things ever. If we limit ourselves to only our internal links or references in the body of an article, that limits us much more.
When someone uses a bit of jargon, or alludes to an argument or concept or debate or body of evidence without wanting to fully explain it there, it seems like good practice to add a link to somewhere where it's explained. (I do this in my regular Forum posts a lot and I advocate for it here.) A substantial portion of the time, the link that's provided is to Wikipedia.
I don't think peppering our wiki articles with inline citations to Wikipedia is a good way to handle this, both because some articles might end up with lots of these in-line citations taking up space and because the term might be somewhat tangential or not the sort of thing where we really want to give a "reference" - we just think some readers might be unfamiliar with the term and want to learn more.
But I'm not 100% sure on this point.
Wikipedia already has an easy way to add footnoted references in which you can place external links, with the text for those references appearing as soon as a reader hovers their cursor over the footnote. This means they can effectively get the convenience-to-a-reader of allowing external links in the body of an article even if they technically disallow it.
To say some overlapping points in another way, I think the three main benefits of allowing external links in the body of an article are:
As noted above, I think sometimes a link to e.g. a Wikipedia article is more appropriate than an in-line citation
E.g. if we just want to provide an optional explanation of a term used in the middle of a sentence.
As a reader, it's more convenient to have the link right there when I'm reading, rather than having to find it in the end section.
This would be solved if we had the same sort of footnote reference thing Wikipedia has, but we don't (yet) have that.
It's lower effort to just add a link than to add an inline citation in a proper format + a full reference in the Bibliography.
So not allowing external links in the body of an article may in effect discourage addition of citations, which seems bad.
Of course, people can just add a link despite what this guide says (or add a rough version of an inline citation + full reference) and let someone else fix it. But overly conscientious people or people worried about their edit being seen as stupid or wrong or bad might (unfortunately) be inclined against doing that.
Given all this, I see four options (there are probably more):
Just allow external links in the body of an article
But this has the downside you mention
Just disallow external links in the body of an article
Just disallow external links in the body of an article, except links to Wikipedia
Arguably, this is "following Wikipedia's policy" to roughly the same extent as option 2 would be
I'm not sure how this scores on the respectability/prestige front, but I'd guess it'd be good?
I think the main thing we want to avoid is having external links to a random assortment of papers and especially blog posts or the like? Having external links to Wikipedia only might be fine? Really not sure.
Do options 2 or 3, but also try to get the same sort of footnote reference feature Wikipedia has as soon as we can
But obviously that's not up to you, and as you say the tech team is very busy
I think maybe the best option is option 3? Not sure, though.
For whatever it's worth, this also seems pretty constraining to me. Internal links are already specially marked via the small degree-symbol, so differentiating internal and external links is pretty straightforward.
Should there be a policy that the "Related entries" section should typically not include links to entries that were already linked to in the body of the entry in question? Or should the "Related entries" section just include links to the most relevant entries, regardless of whether they were already linked to or not?
For example, for the "Dystopia" entry, I think a link to "Totalitarianism" would in theory be worth including in "Related entries", but I'm not sure whether to add it since that entry is already linked to earlier in the "Dystopias" entry.
Initially I was weakly in favor of excluding entries already linked to in the body, but now I'm moderately in favor of including them. My main reason for excluding such entries was that I thought this looked inelegant, but now (1) I'm not sure this is the case and (2) I believe the benefits of having a 'Related entries' section that always lists the most related entries probably outweigh those costs, even if they are genuine. I would be interested in hearing whether you or others agree with this.
You can include things in Related entries even if they were linked to earlier in the article
You're encouraged to do so for the most relevant related entries
For things where it's already not super clear whether they're relevant enough to be in Related entries, you should maybe not include them if they were already linked to earlier (i.e., that could tip the balance towards leaving it out, in ambiguous cases)
But maybe that's unnecessarily complicated. If we had to simplify, I'd weakly vote to keep 1 and 2 and cut 3.
Also, should the Related entries section basically always include a link to the entry for the category/cluster that this entry is part of (where relevant)? For example, should all entries that are part of the Movement building category/cluster include a link to that entry in their Related entries section? And likewise for all entries that are part of the Moral philosophy category/cluster, and all that are part of the Global health & development category/cluster, etc.?
Or is that unnecessary, perhaps because in future users will automatically see where an entry fits in the hierarchy/structure of other entries?
My hunch is that this should be assessed on a case-by-case basis: I would imagine that sometimes it makes sense to include those entries but not on others. I guess the underlying intuition is that the relative degree of relatedness of an entry to its parent isn't constant, and that the variation is such as to justify inclusion in some cases but not in others. But if you find yourself confronting this question repeatedly and feel like it warrants a general policy, let me know (by leaving a comment here).
I think it'd be good to add something similar to Wikipedia's "neutral point of view" policy, covering things like avoiding giving subjective opinions in the EA Wiki's own voice (referring to other people's subjective opinions, such as by quoting them, would still be ok). See discussion here.
I agree that this would be desirable. I have some reservations about Wikipedia's own policy, which places great emphasis on "reliable sources" and understands that notion in a way that would exclude lots of relevant publications (see e.g. Gwern's Wikipedia and dark side editing). I intend to take a look at other existing policies in the coming days, but if you have any concrete suggestions, please let me know.
I think that it makes sense for us to allow or heavily use sources Wikipedia wouldn’t see as reliable. But I think we could still copy other large chunks of their policy, such as on tone?
Okay, I updated the Style Guide with a section on "Tone". Feel free to make any revisions you think are appropriate. I have also edited the Shulman article to remove the problematic sentence.
It looks like Pablo/Aaron/both has been using bold for section headings in entries? Is that something we should all be using? See here for an example where the existing heading was in bold and I added a new heading that isn't: https://forum.effectivealtruism.org/tag/space-colonization
Thanks for flagging this. Section headings should never be boldfaced. I have fixed it in this instance, but would like to identify the origin of the problem, so I can address all entries affected. Off the top of your head, can you recall other articles that exhibit this? I suspect this happened as a result of importing some of the content from EA Concepts.
I can’t remember off the top of my head, but I think it was a substantial portion of the entries I’ve seen and edited. I’ll let you know if I stumble upon more examples.
tl;dr: I think we should use an established citation style (e.g., APA, MLA), rather than creating a new one for this wiki.
(This comment is adapted from a discussion I had with Pablo about an earlier draft of the style guide. I'm sharing it here so other people can upvote, downvote, provide their own thoughts, etc.)
There are already many established citation styles, and they're very widely used, many people are used to them, and many resources on how to use them exist. So it seems to me that one should start off fairly strongly thinking we should just use one of those, and only move from that position if there are quite strong reasons to do so.
I can see two main potential arguments for the EA Wiki having its own, unique citation style:
Perhaps the EA Wiki's needs are very different from the needs of any of the publications, sites, etc. that use established citation styles
(I.e., perhaps those citation styles are good for what they're used for, but wouldn't be ideal here)
Perhaps all of those citation styles are just far from ideal in general, and the one proposed here would just be better in general
At first glance, I'd feel a bit skeptical of either of those arguments.
And even if those arguments true, it's still not obvious that that would make it overall worthwhile to create and use a new citation style. I think the costs of doing that are:
The time spent generating and writing up the basic ideas of this citation style (but this cost has already been paid by Pablo, so it's no longer relevant, really)
The time people will spend reading and later re-checking the citation style section of the EA Wiki style guide, minus the time that would've been spent on that if a standard citation style was suggested instead.
The latter time would likely be smaller because a substantial portion of Forum users are already very familiar with at least one standard format.
The time people will spend editing mistakes that were made in applying this citation style, minus the time people would've spent if a standard citation style was suggested instead.
Again, the latter time would likely be smaller, because:
many Forum users will be better at using existing citation styles due to being more practiced with them
many Forum users will naturally assume they should use existing citation styles (so they'll use them correctly, but that'll be a mistake that needs editing if we've decided they should use a new citation style instead)
many resources exist on existing citation styles, which should make it easier for people to learn to use them; if they don't find one explanation clear or comprehensive, they can find another, and the best will tend to rise to the top
The time people will pend deciding how the Wiki should handle various special cases that may arise that this style guide doesn't get into, or alternatively the lack of consistency that would then occur between different instances of those cases.
Established citation styles involve lots and lots of rules for lots and lots of situations, which seems to suggest any new citation style would naturally tend towards that over time or else would have inconsistencies in how it's used.
Maybe the time people spend debating or explaining the rationale for this new citation style's particular rules or for using it at all rather than established citation styles.
I expect I won't be the only person to raise these sorts of points
And indeed I see evelynciara has already made a comment pushing back on one of the rules in this new citation style
I'm not sure how large those costs are, but I'd guess they're significant, at least relative to the potential benefits of this new citation style relative to using an established citation style.
(Though I can definitely imagine changing my mind about the stuff I've said above.)
Although the Wiki continues to use its own style, we have now added a CSL file which, in combination with a reference management app such as Zotero, allows you to easily add references in the appropriate format, as explained here. If you or anyone else decides to try this, I'd be interested to know if you found it useful.
I have no strong attachment to the current citation format. My main reasons for preferring it are:
I spent a few days, back when I started working on this project, exploring the existing formats (using this tool) and I wasn't able to find a format that handled all the problematic cases in a way I found satisfactory. This was many months ago, so I don't remember the details.
Considerable time has been spent already (mostly by Leo, my excellent assistant) in making sure that all citations conform to the current format. (Most of this time had already been spent by the time you raised this objection a few weeks ago.) The costs are sunk, so that is not in itself a reason, but it provides an estimate of the costs that would have to be incurred to make the citations conform to a new format.
I would be open to the proposal if (1) I was presented with a concrete alternative that handled basic cases correctly (such as not requiring URLs to be listed explicitly) and (2) a quick, back-of-the-envelope calculation of the time that would be saved by adopting this new style. I could then ask Leo to estimate how much time he has spent fixing the citations, and by comparing the two estimates we can decide whether this is worth it.
[I could be wrong about all of the following. Also, this response at least slightly has the vibe of the sort of annoying and counterproductive "slapdown" Eliezer writes about here, partly because I don't have the time to provide a more constructive, detailed, object-level response, so my apologies for any frustration that that causes!]
I spent a few days, back when I started working on this project, exploring the existing formats (using this tool) and I wasn't able to find a format that handled all the problematic cases in a way I found satisfactory. This was many months ago, so I don't remember the details.
As noted above, I see two main potential arguments why that would be true, and I'm a bit skeptical of both for meta reasons. One thing I'd add is that the new citation style has only really been evaluated by its creator (you), I think, so it's possible that part of why it seems better is because of your idiosyncratic views.
But of course neither of us have highlighted specific, object-level arguments (aside from you saying "not requiring URLs to be listed explicitly"), so this is somewhat hard to evaluate. I do think it's plausible that your proposed citation style would be better, and I don't mean to be taken as "slapping down" even the mere idea of trying to generate an alternative to the existing approaches.
One other thing I'd note is that it might be possible to identify an existing citation style that's mostly good but has 5 issues in your view, 3 of which seem especially noteworthy and clearly problematic, and then have our citation style be "that one but with these 3 tweaks". E.g., "APA but you don't need to list URLs explicitly". That might avoid most of the costs I mention from a new citation style and most of the costs you think an existing citation style would have.
Considerable time has been spent already (mostly by Leo, my excellent assistant) in making sure that all citations conform to the current format. (Most of this time had already been spent by the time you raised this objection a few weeks ago.) The costs are sunk, so that is not in itself a reason, but it provides an estimate of the costs that would have to be incurred to make the citations conform to a new format.
At first glance, I think we should probably mostly focus, when making these decisions, on scenarios where the wiki becomes fairly widely used and edited for a long time. Those scenarios seems to account for most of the expected value of work on this project. In those scenarios, there will be much more time spent on entering and editing citations in future than has been spent so far.
So I think the fact that those sunk costs were considerable mostly pushes in favour of thinking carefully about and getting feedback on the decision about citation styles within the next few days / weeks / maybe months. I think it also pushes a bit in favour of keeping the existing citation style, to avoid paying a cost to switch the existing citations to a new style, but I think that that push is probably smaller?
I would be open to the proposal if (1) I was presented with a concrete alternative that handled basic cases correctly (such as not requiring URLs to be listed explicitly) and (2) a quick, back-of-the-envelope calculation of the time that would be saved by adopting this new style. I could then ask Leo to estimate how much time he has spent fixing the citations, and by comparing the two estimates we can decide whether this is worth it.
It does sound like that'd be useful info, but I guess I feel like by default no one will think to and take the time to give you that info even if a different citation style would be better. I don't have time to do this soon myself. (Part of why this wouldn't be super quick is that I don't already have strong views on which citation style would be best; I more so have a meta-level epistemic-humility-style skepticism that any new style that's only really been evaluated by its creator will be better than all existing ones.) So I think I'd say the Forum/wiki team should take it upon themselves to try to work that stuff out or to actively solicit someone else to do so.
In other words: Due to time constraints and lack of detailed cached thoughts on this question, I'm just going for a drive-by "this seems worth thinking about", rather than being able to compelling argue for any particular alternative.
What should our norms about something like "Further reading" links be? I.e., if there's a source that it seems worth pointing readers of an entry to but that isn't referenced in the entry, can/should we include that link in External links? Can/should we have "Further reading" sections just for this?
E.g., on the Accidental harm entry, I think it'd be useful to point readers to one or more of these sources. I could do that by adding text that references these things, but that'd take me longer than just adding the links, and it might just be shoehorny that text in.
(It's possible that this info is already in the style guide and I just missed it. The one relevant thing I saw is that it seems we currently at least have the stance that links to the home page of an org's website are worth adding in External links for the entry on that org.)
ETA: Oh, I now see that there's at least one entry that already has a Further reading section, added by Aaron Gertler: https://forum.effectivealtruism.org/tag/space-colonization So maybe some mention of Further reading sections should be added to the style guide?
The entries with a "Further reading" section are all imported from the EA Concepts. I agree that it would be desirable to have a way of including pointers for further reading, though I am not sure how to best implement this. The problem with a dedicated "Further reading" section is that it's unclear what to do with works deemed worthy of inclusion in that section but which are already included in the Bibliography. I don't think it looks good to list them in both sections, but excluding them is also problematic because the "Further reading" section then effectively becomes something like a section on "all the works the reader may want to consult for more detailed coverage of the subject with the exception of works already listed in the bibliography, some of which are also worth consulting". Essentially, the problem arises because we have two orthogonal distinctions—cited versus not cited, worthy versus not worthy of consultation—that generate four possible combinations, three of which we want to include (the fourth being works neither cited nor worthy of consultation, which of course should not be explicitly listed anywhere), yet we have at most two distinct sections where they could be included. The approach I would tentatively favor is (1) to include works in each of these three categories all under the Bibliography section, which is the practice of the Stanford Encyclopedia of Philosophy and other reference works, and (2) for works worthy of consultation include also a brief, one-sentence descriptive summary, following the practice established by EA Concepts. What do you think of this approach?
Actually, I've just gone on a spree of integrating links to or from my earlier shortform collections into EA Wiki entries, and now I think I see the situation a bit differently. I'd say there are three relevant distinctions, which are neither entirely orthogonal nor entirely the same:
Cited in the entry vs not cited in the entry
Relevant to this topic vs not relevant to this topic
Worth looking at (in general or in relation to this topic), in the editor's subjective opinion
For comparison, the tag system itself seems to be benefitted by a norm of people tagging pretty much any post that seems relevant, unless it really seems quite low quality, rather than only tagging things the tagger themselves thinks are unusually high quality. Then everything relevant is collected in one place, and people can decide what to read for themselves, based on a mixture of how high the tag karma is, how high the regular karma is, the title, the opening paragraphs, the top comment on the post, etc.
But currently there's not necessarily a way to do that for non-Forum sources.
My shortform collections liberally collected anything that seemed relevant (though I sometimes also put things in descending order of usefulness in my subjective opinion), rather than strictly screening for quality. I think this was useful because I was collecting sources on somewhat obscure topics. But this would become ridiculous if done for a topic like "Existential risk".
And even if it didn't lead to a ridiculous number of things mentioned, I feel like inclusion of something in an EA Wiki Further reading or External links section would by default carry a stronger implication that this is high quality than just me including it in a personally made (and slightly messy) shortform would?
So maybe we need a norm about how notable or perceived-as-high-quality-by-us an external link should be for us to mention it in these sections of Wiki entries. But I'm not sure what the norm should be. And maybe it'd be better to have an additional type of section or a linked google doc or something where people can more liberally mention anything that's relevant, and this doesn't clog up the main part of the entry itself, and it's more obvious that these sources haven't been screened for quality?
(Also, FWIW, this morning I just fairly randomly used either Further reading, Bibliography, or External links for mentioning relevant external works that weren't cited in the article, partly depending on what section was already in the entry. This is a bit messy, but I think it was better to do and then let it be cleaned up later than not to add those things at all.)
Sorry for the delay in responding—currently I do not get a notification when someone comments on an entry.
My impression is that it is not uncommon for a citation style to omit a punctuation mark when the author's name is followed by the date of publication enclosed in parenthesis. The first Google Images search I ran produced this; I don't know who popular BIOI is, but at least it establishes that this practice has some precedent. My approach has generally been to omit punctuation marks in citations unless necessary (that's why there is no comma between the author's surname and the year in inline citations), so I would prefer to maintain the current format. But feel free to raise anything else that strikes you as worthy of flagging.
The Tags Portal page's link to the style guide actually links to a doc that then itself links to the style guide. Could the portal be edited to link directly to the style guide?
I would suggest that we don't encourage citations to books to include the name or (especially) city of the publisher. This information is almost always superfluous.
Yeah, I agree it's superfluous. I was reminded of this tweet by Diana Fleischman. Scholarly conventions change at a glacial pace, but unfortunately this convention is followed by all the major reference works that I'm aware of, including Wikipedia. Since the formatting can be outsourced to an assistant, I think the costs of adhering to this convention are worth its relatively low costs.
Should entry/tag names use "&" or "and"? E.g., should it be "sentience & consciousness" or "sentience and consciousness"?
It seems like recent entries/tags - since the announcement of the wiki, not just tags - are mostly using "and". And maybe that looks more encyclopaedic. So I guess we should go with that?
Yes, they should use "and". I updated the Guide to reflect this.
There are many entries/tags for which there is a very related EA-aligned Facebook group (see here for a directory to such groups). I think that, in all such cases, it'd probably be good to add a link to the relevant group in an External links section in the entry/tag. (This is for the obvious reason that many people checking out the entry/tag might also find the Facebook group useful, e.g. to find more info, get career advice, or share relevant drafts for feedback.)
Before I go adding all those links, does anyone think that that's a bad idea?
As for how related the Facebook group should be in order for it to be linked to, the rule of thumb I'm leaning towards is "Each Facebook group should only be linked to from one entry/tag. If the Facebook group seems relevant to multiple entries/tags, that suggests that those entries/tags should probably link to each other in the Related entries section, and the Facebook group should just be linked to from the most relevant one. This should still allow the readers most likely to be interested in the group to find it."
E.g., I'd link to the Risks of astronomical suffering (s-risks) Facebook group from the s-risks entry; but I wouldn't link to it from the entries on suffering and pain, existential risks, trajectory changes, the long-term future, mind crime, etc.; but I'd ensure many of the latter entries link to the s-risks entry.
(I also plan to post in the relevant Facebook groups to mention that the relevant tags/entries exist, so people can check out the posts, tag more posts, or edit the entries [though I'll skip groups where I did this already last year or this year]. But I feel more confident that that makes sense to do, and it's not really a matter for this style guide anyway since it involves promoting rather than editing the entries.)
I don't a strong opinion either way. Part of me wants to encourage people to use the Forum to conduct these discussions, rather than point them to a group on a closed social media platform with mediocre editing and commenting features, especially insofar as this to a significant extent happens because of network effects and trivial inconveniences (rather than revealing an inherent preference for one venue over the other). I also worry slightly that linking to Facebook may make the entries look a bit amateurish and not intellectually serious. On the other hand, I recognize that such links could help users find relevant content and connect them to others with similar interests. I'd like to know what others think before we make a general decision on this.
(By the way, I'm using 'entry' as a neutral term to refer to what may, depending on what functionality is emphasized, be either called a 'tag' or a 'wiki article'. We may want to use that term in the future.)
What should our norms be, or what is the legal situation with, copying text from LessWrong wiki entries verbatim as the text for a Forum wiki entry?
I guess we should probably paraphrase and cite them, rather than copying verbatim, but maybe we don't have to and sometimes it's better (to save time) to at least allow a verbatim copy?
Should there be a norm where, for (almost?) every Forum wiki entry that has a corresponding Wikipedia article, the Further reading / Bibliography / External links section includes a link to that Wikipedia article?
I'd be opposed to such a norm. Very often, Wikipedia is not the best reference on a given topic, and their articles are already extremely easy to find. I would decide whether to cite them on a case by case basis, with a relatively high bar for citing them.
(A rule of this sort may be more plausible with reference works of exceptional quality, such as the Stanford Encyclopedia of Philosophy, but even then it doesn't feel to me like we should have such a rule. I guess the underlying intuition is that I don't see a reason to deviate from the general principle that decisions on whether to include a particular work in the bibliography should be based on an assessment of that specific work's quality and relevance, rather than on some general rule.)
These are good points I hadn't really considered.
I think what I had in mind was something like this: In cases where the topic covered by a Forum wiki entry is also covered by a Wikipedia article or an SEP article or whatever:
(I wrote that quickly; not sure I explained it well.)
But, as mentioned, I think your points are good too, so now I'm unsure what norm I'd favour here.
I very much agree with your points, especially the first one. Perhaps as we gain more experience, we can get a better sense of the types of articles that warrant a policy of linking to some external source by default. I can imagine that being the case e.g. for core philosophy topics (e.g. 'normative ethics') and the Stanford Encyclopedia of Philosophy.
And what should the norm be for cases where there's a corresponding LessWrong wiki entry / concept / tag? So far, I've been including a link to that in the Further reading section in (almost?) any case where the Forum wiki entry has little text and/or the LessWrong entry has a bunch of text. But it's plausible I should do this more or less than I currently do.
The principle I've been following is to treat LW like I do all other sources, and cite their articles iff they seem worthy of inclusion. (I do think it's always worth checking out the LW tags, because they are more likely to pass that test than most other sources.)
Sounds good to me! (Also sounds very obvious in retrospect, I guess.)
Is this style guide the right place for policies/norms about how to use tags? E.g., a policy about which posts should be tagged with a tag for an organisation, as discussed here?
Or is there/should there be some other place for such policies/norms?
Seems like that's more about "tagging" and less about "style for the wiki entries".
This doesn't seem like the right place. I've written my thoughts on tagging norms in a few places, but it would be good to collect that (plus others' thoughts) in one place. I'll talk to Pablo and see about one of us creating this resource.
Meanwhile, if you have thoughts on tagging policy, feel free to mention them in replies to this comment (or to create a question post to collect others' thoughts — though I'm not sure how much discussion is required here, as I think our tagging norms/policy will end up being pretty simple).
Update: We discussed this with Aaron and we'll put up a document shortly.
I think the Style Guide is not the right place for discussion of these norms. As you note, this concerns tagging rather than Wiki content. I'd tentatively suggest commenting on the proposed tags thread, but I'll ask Aaron.
I think external links should be allowed in the lead or body sections. And in discussions on a draft of this style guide, I seem to recall that you (Pablo) moved to agreeing with that view? (ETA: As discussed in replies, it turns out I misremembered that.) Should that line of the style guide be changed?
(Let me know if I should explain my reasoning for this position again.)
I don't recall agreeing to this. As I wrote in a comment to the Google Doc:
The approaches alluded there are to allow users to insert citations entries from a database rather than having to enter the bibliographic details manually. This should make the insertion of a citation roughly as easy as the insertion of a link, so requiring one over the other shouldn't feel constraining. I understand that the tech team is very busy with other stuff, so I'm not sure whether we can expect any of these innovations to be developed soon (or at all), but for the time being this work can be delegated to our assistant, and we will make it clear in our FAQ users should not waste any time formatting citations. If the situation doesn't change in 3-6 months, we can reconsider.
I realize I probably haven't made this very clear, but a core principle I've been following when writing content and planning (and one that I feel reluctant to abandon, given its centrality to how this project was conceived) is to follow existing scholarly norms as closely as reasonably possible. I believe this is generally desirable insofar as many of these norms can be observed at little or no cost, while providing significant benefits in the form of recognition and authoritativeness, but is especially important for a Wiki that will cover areas that are in many ways speculative and are either neglected or not held in especially high esteem by academia, journalism, and other elite institutions.
So my opposition to having external links in the body of an entry is based on my perception that this is generally not an accepted scholarly practice. Wikipedia doesn't normally allow the use of such links except in special sections (such as "external links" and references sections), and my casual acquaintance with other encyclopedias and reference works suggests that they too follow norms of this sort. If you or others could point me to counterexamples sufficient to establish that this is not in fact common, I would be happy to revise the policy, to which I have no personal attachment.
Ah, yes, I just checked the doc comments and you're right. I must've simply misremembered.
That makes sense, and I think I'd underestimated that argument in favour of not allowing external links. So I'm now less sure what we should do here.
I don't have any counterexamples. Instead, I'll point to some reasons why I think allowing external links may make more sense for our wiki than for Wikipedia:
To say some overlapping points in another way, I think the three main benefits of allowing external links in the body of an article are:
Given all this, I see four options (there are probably more):
I think maybe the best option is option 3? Not sure, though.
For whatever it's worth, this also seems pretty constraining to me. Internal links are already specially marked via the small degree-symbol, so differentiating internal and external links is pretty straightforward.
Should there be a policy that the "Related entries" section should typically not include links to entries that were already linked to in the body of the entry in question? Or should the "Related entries" section just include links to the most relevant entries, regardless of whether they were already linked to or not?
For example, for the "Dystopia" entry, I think a link to "Totalitarianism" would in theory be worth including in "Related entries", but I'm not sure whether to add it since that entry is already linked to earlier in the "Dystopias" entry.
Initially I was weakly in favor of excluding entries already linked to in the body, but now I'm moderately in favor of including them. My main reason for excluding such entries was that I thought this looked inelegant, but now (1) I'm not sure this is the case and (2) I believe the benefits of having a 'Related entries' section that always lists the most related entries probably outweigh those costs, even if they are genuine. I would be interested in hearing whether you or others agree with this.
I think I'd weakly vote for a policy like:
But maybe that's unnecessarily complicated. If we had to simplify, I'd weakly vote to keep 1 and 2 and cut 3.
Also, should the Related entries section basically always include a link to the entry for the category/cluster that this entry is part of (where relevant)? For example, should all entries that are part of the Movement building category/cluster include a link to that entry in their Related entries section? And likewise for all entries that are part of the Moral philosophy category/cluster, and all that are part of the Global health & development category/cluster, etc.?
Or is that unnecessary, perhaps because in future users will automatically see where an entry fits in the hierarchy/structure of other entries?
My hunch is that this should be assessed on a case-by-case basis: I would imagine that sometimes it makes sense to include those entries but not on others. I guess the underlying intuition is that the relative degree of relatedness of an entry to its parent isn't constant, and that the variation is such as to justify inclusion in some cases but not in others. But if you find yourself confronting this question repeatedly and feel like it warrants a general policy, let me know (by leaving a comment here).
Just to flag that I was not notified of all these comments, so I see them only now. I will start responding shortly.
I think it'd be good to add something similar to Wikipedia's "neutral point of view" policy, covering things like avoiding giving subjective opinions in the EA Wiki's own voice (referring to other people's subjective opinions, such as by quoting them, would still be ok). See discussion here.
I agree that this would be desirable. I have some reservations about Wikipedia's own policy, which places great emphasis on "reliable sources" and understands that notion in a way that would exclude lots of relevant publications (see e.g. Gwern's Wikipedia and dark side editing). I intend to take a look at other existing policies in the coming days, but if you have any concrete suggestions, please let me know.
I think that it makes sense for us to allow or heavily use sources Wikipedia wouldn’t see as reliable. But I think we could still copy other large chunks of their policy, such as on tone?
Okay, I updated the Style Guide with a section on "Tone". Feel free to make any revisions you think are appropriate. I have also edited the Shulman article to remove the problematic sentence.
The policy, and the tweaked Shulman article, both look good to me!
(And thanks again for all your work on the Wiki - I'm really excited about the shape it's taking.)
Yes, I agree. I'll try to draft something later today.
It looks like Pablo/Aaron/both has been using bold for section headings in entries? Is that something we should all be using? See here for an example where the existing heading was in bold and I added a new heading that isn't: https://forum.effectivealtruism.org/tag/space-colonization
Thanks for flagging this. Section headings should never be boldfaced. I have fixed it in this instance, but would like to identify the origin of the problem, so I can address all entries affected. Off the top of your head, can you recall other articles that exhibit this? I suspect this happened as a result of importing some of the content from EA Concepts.
I can’t remember off the top of my head, but I think it was a substantial portion of the entries I’ve seen and edited. I’ll let you know if I stumble upon more examples.
I've now stumbled upon some examples:
(I've now edited them to remove bolded section names, but you could still check their edit history.)
Thanks, yeah, it looks this was related to EA Concepts. I've made a note to be on the look for other instances of this and fix them when I find them.
tl;dr: I think we should use an established citation style (e.g., APA, MLA), rather than creating a new one for this wiki.
(This comment is adapted from a discussion I had with Pablo about an earlier draft of the style guide. I'm sharing it here so other people can upvote, downvote, provide their own thoughts, etc.)
There are already many established citation styles, and they're very widely used, many people are used to them, and many resources on how to use them exist. So it seems to me that one should start off fairly strongly thinking we should just use one of those, and only move from that position if there are quite strong reasons to do so.
I can see two main potential arguments for the EA Wiki having its own, unique citation style:
At first glance, I'd feel a bit skeptical of either of those arguments.
And even if those arguments true, it's still not obvious that that would make it overall worthwhile to create and use a new citation style. I think the costs of doing that are:
I'm not sure how large those costs are, but I'd guess they're significant, at least relative to the potential benefits of this new citation style relative to using an established citation style.
(Though I can definitely imagine changing my mind about the stuff I've said above.)
Hi Michael,
Although the Wiki continues to use its own style, we have now added a CSL file which, in combination with a reference management app such as Zotero, allows you to easily add references in the appropriate format, as explained here. If you or anyone else decides to try this, I'd be interested to know if you found it useful.
Thanks for the detailed response.
I have no strong attachment to the current citation format. My main reasons for preferring it are:
I would be open to the proposal if (1) I was presented with a concrete alternative that handled basic cases correctly (such as not requiring URLs to be listed explicitly) and (2) a quick, back-of-the-envelope calculation of the time that would be saved by adopting this new style. I could then ask Leo to estimate how much time he has spent fixing the citations, and by comparing the two estimates we can decide whether this is worth it.
That reply seems reasonable.
[I could be wrong about all of the following. Also, this response at least slightly has the vibe of the sort of annoying and counterproductive "slapdown" Eliezer writes about here, partly because I don't have the time to provide a more constructive, detailed, object-level response, so my apologies for any frustration that that causes!]
As noted above, I see two main potential arguments why that would be true, and I'm a bit skeptical of both for meta reasons. One thing I'd add is that the new citation style has only really been evaluated by its creator (you), I think, so it's possible that part of why it seems better is because of your idiosyncratic views.
But of course neither of us have highlighted specific, object-level arguments (aside from you saying "not requiring URLs to be listed explicitly"), so this is somewhat hard to evaluate. I do think it's plausible that your proposed citation style would be better, and I don't mean to be taken as "slapping down" even the mere idea of trying to generate an alternative to the existing approaches.
One other thing I'd note is that it might be possible to identify an existing citation style that's mostly good but has 5 issues in your view, 3 of which seem especially noteworthy and clearly problematic, and then have our citation style be "that one but with these 3 tweaks". E.g., "APA but you don't need to list URLs explicitly". That might avoid most of the costs I mention from a new citation style and most of the costs you think an existing citation style would have.
At first glance, I think we should probably mostly focus, when making these decisions, on scenarios where the wiki becomes fairly widely used and edited for a long time. Those scenarios seems to account for most of the expected value of work on this project. In those scenarios, there will be much more time spent on entering and editing citations in future than has been spent so far.
So I think the fact that those sunk costs were considerable mostly pushes in favour of thinking carefully about and getting feedback on the decision about citation styles within the next few days / weeks / maybe months. I think it also pushes a bit in favour of keeping the existing citation style, to avoid paying a cost to switch the existing citations to a new style, but I think that that push is probably smaller?
It does sound like that'd be useful info, but I guess I feel like by default no one will think to and take the time to give you that info even if a different citation style would be better. I don't have time to do this soon myself. (Part of why this wouldn't be super quick is that I don't already have strong views on which citation style would be best; I more so have a meta-level epistemic-humility-style skepticism that any new style that's only really been evaluated by its creator will be better than all existing ones.) So I think I'd say the Forum/wiki team should take it upon themselves to try to work that stuff out or to actively solicit someone else to do so.
In other words: Due to time constraints and lack of detailed cached thoughts on this question, I'm just going for a drive-by "this seems worth thinking about", rather than being able to compelling argue for any particular alternative.
What should our norms about something like "Further reading" links be? I.e., if there's a source that it seems worth pointing readers of an entry to but that isn't referenced in the entry, can/should we include that link in External links? Can/should we have "Further reading" sections just for this?
E.g., on the Accidental harm entry, I think it'd be useful to point readers to one or more of these sources. I could do that by adding text that references these things, but that'd take me longer than just adding the links, and it might just be shoehorny that text in.
(It's possible that this info is already in the style guide and I just missed it. The one relevant thing I saw is that it seems we currently at least have the stance that links to the home page of an org's website are worth adding in External links for the entry on that org.)
ETA: Oh, I now see that there's at least one entry that already has a Further reading section, added by Aaron Gertler: https://forum.effectivealtruism.org/tag/space-colonization So maybe some mention of Further reading sections should be added to the style guide?
The entries with a "Further reading" section are all imported from the EA Concepts. I agree that it would be desirable to have a way of including pointers for further reading, though I am not sure how to best implement this. The problem with a dedicated "Further reading" section is that it's unclear what to do with works deemed worthy of inclusion in that section but which are already included in the Bibliography. I don't think it looks good to list them in both sections, but excluding them is also problematic because the "Further reading" section then effectively becomes something like a section on "all the works the reader may want to consult for more detailed coverage of the subject with the exception of works already listed in the bibliography, some of which are also worth consulting". Essentially, the problem arises because we have two orthogonal distinctions—cited versus not cited, worthy versus not worthy of consultation—that generate four possible combinations, three of which we want to include (the fourth being works neither cited nor worthy of consultation, which of course should not be explicitly listed anywhere), yet we have at most two distinct sections where they could be included. The approach I would tentatively favor is (1) to include works in each of these three categories all under the Bibliography section, which is the practice of the Stanford Encyclopedia of Philosophy and other reference works, and (2) for works worthy of consultation include also a brief, one-sentence descriptive summary, following the practice established by EA Concepts. What do you think of this approach?
Actually, I've just gone on a spree of integrating links to or from my earlier shortform collections into EA Wiki entries, and now I think I see the situation a bit differently. I'd say there are three relevant distinctions, which are neither entirely orthogonal nor entirely the same:
For comparison, the tag system itself seems to be benefitted by a norm of people tagging pretty much any post that seems relevant, unless it really seems quite low quality, rather than only tagging things the tagger themselves thinks are unusually high quality. Then everything relevant is collected in one place, and people can decide what to read for themselves, based on a mixture of how high the tag karma is, how high the regular karma is, the title, the opening paragraphs, the top comment on the post, etc.
But currently there's not necessarily a way to do that for non-Forum sources.
My shortform collections liberally collected anything that seemed relevant (though I sometimes also put things in descending order of usefulness in my subjective opinion), rather than strictly screening for quality. I think this was useful because I was collecting sources on somewhat obscure topics. But this would become ridiculous if done for a topic like "Existential risk".
And even if it didn't lead to a ridiculous number of things mentioned, I feel like inclusion of something in an EA Wiki Further reading or External links section would by default carry a stronger implication that this is high quality than just me including it in a personally made (and slightly messy) shortform would?
So maybe we need a norm about how notable or perceived-as-high-quality-by-us an external link should be for us to mention it in these sections of Wiki entries. But I'm not sure what the norm should be. And maybe it'd be better to have an additional type of section or a linked google doc or something where people can more liberally mention anything that's relevant, and this doesn't clog up the main part of the entry itself, and it's more obvious that these sources haven't been screened for quality?
(Also, FWIW, this morning I just fairly randomly used either Further reading, Bibliography, or External links for mentioning relevant external works that weren't cited in the article, partly depending on what section was already in the entry. This is a bit messy, but I think it was better to do and then let it be cleaned up later than not to add those things at all.)
I think it's weird not to have punctuation between the author and title, as in this example:
Pretty much all major citation styles (e.g. MLA and APA) have a period after the author's name.
Sorry for the delay in responding—currently I do not get a notification when someone comments on an entry.
My impression is that it is not uncommon for a citation style to omit a punctuation mark when the author's name is followed by the date of publication enclosed in parenthesis. The first Google Images search I ran produced this; I don't know who popular BIOI is, but at least it establishes that this practice has some precedent. My approach has generally been to omit punctuation marks in citations unless necessary (that's why there is no comma between the author's surname and the year in inline citations), so I would prefer to maintain the current format. But feel free to raise anything else that strikes you as worthy of flagging.
The Tags Portal page's link to the style guide actually links to a doc that then itself links to the style guide. Could the portal be edited to link directly to the style guide?
Thanks, I don't have access to the Tags Portal but I asked Aaron to update the link.
Citation Formatting
I would suggest that we don't encourage citations to books to include the name or (especially) city of the publisher. This information is almost always superfluous.
Yeah, I agree it's superfluous. I was reminded of this tweet by Diana Fleischman. Scholarly conventions change at a glacial pace, but unfortunately this convention is followed by all the major reference works that I'm aware of, including Wikipedia. Since the formatting can be outsourced to an assistant, I think the costs of adhering to this convention are worth its relatively low costs.