Should we use wiki to improve knowledge management within the community?

post by VPetukhov · 2019-12-08T23:30:48.143Z · score: 31 (14 votes) · EA · GW · 25 comments

This is a question post.

Contents

  Problem
  Motivation
None
25 comments

TLDR: is there any reason EA community doesn't actively use wiki approach? Or do I miss something and it does?

Problem

Searching for "effective altruism wiki" I mostly meet broken links. Several years ago the announcement [EA · GW] was posted that wiki moves to EA Hub. None of the three links posted there works, but EA Hub has Priority Wiki. However Priority Wiki is focused only on cause prioritization and is almost empty know. So I'm wondering, what happened to the old EA Wiki? Were there any reasons for closing it? What happened to the knowledge published there? And, more broadly, why creating a common knowledge base is almost (see below) never mentioned as an impactful contribution?(see below)

Motivation

During the last half year I was talking a lot with people exploring different EA options (like me), and each of them performed some EA-related research and shared some valuable information with me. So I'm surprised why such kind of knowledge isn't gathered together. Even gathering information about all organizations related to some topic is troubling: there are lots of them with different quality and roles. Definitely, 80k does great job collecting it in the problem profiles. But wouldn't it be more efficient (and effective) to outsource this task to the community? Another example: just today I met two old EA lists (1 [EA(p) · GW(p)], 2 [EA · GW]), which would be nice to have updated and open for contribution.

Indeed, I've found that several years ago Vipul Naik was contributing actively [EA · GW] to EA-related articles on wikipedia. And according to his page, he switched to timelines, which are mostly focused on EA. Moreover, Foundational Research Institute also proposes contributing to wikipedia as one of the volunteering activities.

But with all my respect to wikipedia, I think that having a local wiki would allow to focus on more action-related topics instead of some general knowledge. And it would greatly simplify searching for such information and increase chances that the content you write will be read by people with similar goals.

Answers

25 comments

Comments sorted by top scores.

comment by EdoArad (edoarad) · 2019-12-09T08:23:19.961Z · score: 22 (9 votes) · EA(p) · GW(p)

Some thoughts:

In summary, the empirical results paint a somewhat different picture of sustained contribution than originally hypothesized. Specifically, sustained contributors appear to be motivated by a perception that the project needs their contributions (H1); by abilities other than domain expertise (H2); by personal rather than social motives (H3 & 4); and by intrinsic enjoyment of the process of contributing (H7) rather than extrinsic factors such as learning (H6). In contrast, meta-contributors seem to be motivated by social factors (H3 & 4), as well as by intrinsic enjoyment (H7).

  • I think that we should strategically plan how to incentivize possible contributors. Ideally, people should contribute based on what would be the most valuable, which is something that may be achievable through prizes (possibly "Karma" or money, but perhaps better is something like certificates of impact [EA · GW]), bounties, peer support and acknowledgment, and requests and recognition from leaders of the community.
  • I think that it would take a big effort to bootstrap something new. The efforts going into EA Hub seems to me like a good place to start a centralized knowledge base.
  • I'd like something like a top/bottom research agenda on "how to do the most good", that ends with concrete problems ([like these])(https://forum.effectivealtruism.org/posts/2zcBy7eDXjEti9Sw7/a-collection-of-researchy-projects-for-aspiring-eas [EA · GW]). Something that can help us be more strategic in our resource allocation, and through which we can more easily focus experts on where they can help the most (and have a good infrastructure for moral trade).
  • It seems that something like Roam could be great, because it is designed to make it easy to create pages and has backlinks to support exploration and has other neat stuff. It is still not mature enough though.
comment by vaidehi_agarwalla · 2019-12-10T01:04:37.615Z · score: 8 (3 votes) · EA(p) · GW(p)

I think collectiven intelligence could be a really promising task Y [EA · GW] if the basic structure is designed properly. I'd guess the biggest bottleneck here is funding to build the infrastructure.

In contrast, meta-contributors seem to be motivated by social factors (H3 & 4), as well as by intrinsic enjoyment (H7).

The differences between meta-contributors and contributors is probably really important to understand, and may have implications for a lot of different volunteer projects - one failure mode could be if meta-contributors setting up a project and then failing to find the right people to contribute (i.e. finding people like themselves who are motivated by social rather than personal factors)

Thanks for the research links, and the summary!

comment by EdoArad (edoarad) · 2019-12-10T06:14:56.600Z · score: 3 (2 votes) · EA(p) · GW(p)

I'm surprised that you think that the bottleneck is in funding, I guess that means that I overestimate the easiness and desirability of using some existing tools.
Interested in your take on it :)

comment by vaidehi_agarwalla · 2019-12-10T11:45:49.907Z · score: 1 (1 votes) · EA(p) · GW(p)

Perhaps that is misleading - let me rephrase - if the solution really requires creating a new infrastructure then I'd say funding would be a bottle neck (not that funds don't exist, but they aren't always easy to get), but getting the right people to build the infrastructure would probably be a bottleneck too.

What existing tools were you thinking of?

comment by EdoArad (edoarad) · 2019-12-10T17:07:15.618Z · score: 2 (2 votes) · EA(p) · GW(p)

I actually did not give that enough thought. I think using MediaWiki or Wikidot might be fine for start, and I am very fond of Roam. Notion might be great here as well. All of them require getting used to because the syntax is not straightforward, but that suffices for textual edits if there are people who go over and fix design problems. Roam is more difficult because it is... different.. and because it is less mature. Roam being in it's starting phases might actually be a good thing, because it's development can probably shift to the needs of the EA community in this case if the EA Wiki will be hosted there (Roam Research received a grant from the Long Term Future Fund [EA · GW])

That is all to say that I think a basic wiki infrastructure might be fine for start, if there is a good roadmap and support from the community. I assume that markets and fancy prizes can wait for later or be hacked into existence, but maybe that should be in the design from the start 🤷‍♂️

comment by VPetukhov · 2019-12-09T22:57:28.354Z · score: 2 (2 votes) · EA(p) · GW(p)
This post on how to find EA documents [EA · GW], the forum pingbacka [EA · GW] and asking questions on the forum are some examples of ways to find information.

That's very useful, thanks! Pingbacks especially.

There is research on what gets people motivated to write in wikipedia. Here is a recent study that found some interesting stuff, and they conclude with

Wow, so we even have some theory on that. And motivation from the paper looks aligned with EA values.

I think that we should strategically plan how to incentivize possible contributors. Ideally, people should contribute based on what would be the most valuable, which is something that may be achievable through prizes

Completely agree. And also really appreciate your science-based approach. We definitely should discuss it if more of us agree that some platform for open contributions is needed.

I think that it would take a big effort to bootstrap something new. The efforts going into EA Hub seems to me like a good place to start a centralized knowledge base.

+1

I'd like something like a top/bottom research agenda on "how to do the most good", that ends with concrete problems

Yep! We also have such list in EA Denmark though ours is much simpler. And indeed, it was one of the things that pushed me to ask this question.

It seems that something like Roam could be great, because it is designed to make it easy to create pages and has backlinks to support exploration and has other neat stuff. It is still not mature enough though.

I was thinking a lot about GitHub-like structures. It's too complicated for general knowledge, but designed pretty well for more complex domains. For example, if a group works on something like OpenPhil Cause Reports, where producing each piece of information takes long time, and also work of the beginners must be validated by more experienced users. In such cases system of branches allows splitting publishing-ready information from work in progress, issues allow to contribute for those who isn't skilled enough to create product, but has enough experience to note a problem. And so on. But that's just one of possibilities.

This can be possibly implemented adequately on the forum (but requires better search, better norm for writing information, and a better norm of referencing to other materials, perhaps in the comments).

Not sure about that. Different kind of information requires different types of knowledge bases. And here as an example we can take forums, StackOverflow and wikipedia. As far as I understand, if you want to share information about a topic, where some consensus can be found over time and once found is not expected to be changed quickly, then you want wiki. So, someone writes an article with main ideas about the topic and others polish it up (which is kind of impossible on forums). Another situation is if your domain changes too quickly (such as programming languages). Then there is no reason in having overhead for having nicely written articles about every aspect of it. At maximum you'll need to have some blog posts. And the third popular case is asking for personal opinion: it can be either some tips and hacks (StackOverflow has plenty) or just discussion of some ideas like we do know. And this is impossible on wiki.

My impression is that in EA community we lack well-organized up-to-date information, which would represent some kind of consensus instead of a bunch of personal opinions. Your list "how to do the most good" is one example of a thing, which can't be implemented on forum. For such lists, suggestions to EA Hub resources, which @cafelow mentioned would solve the problem. But in general I wouldn't expect them being as effective as wiki.

And this is an interesting experiment in a mechanism designed to improve incentives for collective knowledge production.

Hah, if you have problems with incentives, just add some markets! :)

comment by EdoArad (edoarad) · 2019-12-10T06:36:26.408Z · score: 2 (2 votes) · EA(p) · GW(p)

Re Github-like structures, I think that Google Docs can be sufficient for most cases. Instead of branches, you have non-published docs. And using a wiki page instead of issues might be fine.

I agree with your analysis of knowledge bases, thanks for clarifying that! I take back the suggestion of doubling down on the forum mostly because it seems difficult to properly keep the information updated and to have a clear consensus.

comment by EdoArad (edoarad) · 2019-12-09T08:52:24.700Z · score: 2 (2 votes) · EA(p) · GW(p)
  • Also, I found that I tend to access Wikipedia mostly as a search result, and sometimes go deeper if there are inner links that interest me. This means that we only need the information to be accessible by search, and to be good at referencing further material. This can be possibly implemented adequately on the forum (but requires better search, better norm for writing information, and a better norm of referencing to other materials, perhaps in the comments).

  • And this is an interesting experiment in a mechanism designed to improve incentives for collective knowledge production.

comment by vaidehi_agarwalla · 2019-12-10T01:04:33.698Z · score: 1 (1 votes) · EA(p) · GW(p)

(accidentally commented twice)

comment by saulius · 2019-12-10T12:14:18.028Z · score: 14 (9 votes) · EA(p) · GW(p)

Similarly to EdoArad, I am afraid that without some incentive/gamification structure, there would be a lack of motivation for users to edit an EA wiki. I haven’t read any science on motivations but I want to share my personal point of view. Personally, I don’t feel very motivated to edit the priority wiki because:

  • I will probably receive no comments or feedback on what I wrote
  • I don’t know who and when will read whatever I write there
  • Even if someone reads it, it’s unlikely that they will notice that it was me who wrote it. That means that writing there won’t improve my career prospects or social status.[1] No one will come to me at the EAG or something to talk about this thing that I wrote.

Note that none of these problems apply to the EA forum which is why it feels much more motivating to write here.

I used to edit Lithuanian Wikipedia and these things didn’t apply there either because:

  • I knew that many people were reading pages I was editing. It was very useful and motivating to check pageview statistics.
  • There was a community, editing wikipedia felt like a collaborative project because:
    • There were often discussions on talk pages and a skype channel
    • I knew that admins and some other users on the Lithuanian wikipedia are looking at the recent changes page (which is much less eventful for smaller wikis) and that my new contributions will not go unnoticed. It was irrational for me to try to impress these people whom I have never met in real life but the human brain is wired to care about this kind of stuff (or at least my brain).
    • People would give other people “awards” which were just pieces of html that you could put on your user page. E.g. see awards for this random wikipedia user. Receiving such awards made me feel like my effort was appreciated.
    • We also had an “article of the week” which was a new article written that week that would be featured on the frontpage. I often tried to get my article featured and then it would get a lot of views. It’s almost like gamifying contributions
    • Best articles on wikipedia are marked as featured or good. It was motivating to try to get my articles one of these statuses.

I’m not sure if I’m suggesting to have any of these motivation structures for an EA wiki though. It would probably not reach the critical mass where doing some of this stuff would start making sense. If it did reach the critical mass, I’d be afraid to put too many of these motivation structures in place. We don’t want EAs are spending too much time editing an EA wiki instead of doing more direct things to help the world. But maybe there is some middle ground here.


  1. It’s probably obvious but I feel like I should clarify here that I’m not saying that I contribute to the EA forum only to get social status, etc. I do want to make an impact. But it’s difficult to motivate myself every day just by the thought that what I do might make an impact. Hence it’s good to put myself in situations where I care to do the same things that make an impact for other reasons. It’s like going to an exercise class because you know that you will be ashamed to not do exercise there when everyone around you is doing it. It’s not like you do exercise to impress those strangers, you do it to get fit, but it’s difficult to motivate yourself by the thought of getting fit alone. ↩︎

comment by VPetukhov · 2019-12-14T12:07:21.527Z · score: 1 (1 votes) · EA(p) · GW(p)

I'm wondering if we can address the problem with the same framework as for having better list of concrete projects [EA · GW]? Let's say, we have a list of articles suggested for contribution, like this one, but maybe with some additional info on prioritization. A person picks the article the same way as a project and mark it somehow as "Taken" and then "Done". So we can trace their contribution based on this selection process and assign status points correspondingly. So, essentially, "Writing article" is just a possible type of a project, and the rest workflows are the same. Which suggests that we need to think more about project-based contribution. Perhaps it worth referencing Effective Thesis here as a similar initiative.

And having something slack for wikipedia contributors doesn't seem like a problem at all.

Alternative approach could be adopting MediaWiki and integrating it with existing user rating system. I don't know how backend for EA Forum is organized, but something like OpenID could be a relatively painless solution.

comment by cafelow · 2019-12-09T04:53:03.408Z · score: 8 (7 votes) · EA(p) · GW(p)

Re: The old wiki on the EA Hub, I'm afraid the old wiki data got corrupted, it wasn't backed up properly and it was deemed too difficult to restore at the time :(. So it looks like the information in that wiki is now lost to the winds. I'm very sorry about that.

This may or may not fulfil your needs (and isn't _quite_ a wiki), but the EA Hub resources (resources.eahub.org), is a repository for EA links, and we hope to grow the number of resources available. We will soon have a way for people to make suggestions for changes and additions. which will be vetted before loading onto the website.


comment by riceissa · 2019-12-09T09:49:46.386Z · score: 7 (5 votes) · EA(p) · GW(p)

Re: The old wiki on the EA Hub, I'm afraid the old wiki data got corrupted, it wasn't backed up properly and it was deemed too difficult to restore at the time :(. So it looks like the information in that wiki is now lost to the winds.

I think a dump of the wiki is available at https://archive.org/details/wiki-wikieahuborg_w.

comment by VPetukhov · 2019-12-09T21:47:37.191Z · score: 1 (1 votes) · EA(p) · GW(p)

Unfortunately, no. The archive there contains only the html with the main page and some logos...

comment by JimmyJ · 2019-12-10T00:20:19.236Z · score: 13 (9 votes) · EA(p) · GW(p)

Hi, as the person who personally generated the wiki dump, I can assure you that the complete content of every edit revision of every article was saved, and the data is saved in an XML format that can be trivially imported into MediaWiki. Additionally, I grabbed it after site activity had already died down, but before the wiki got taken over by spambots, so the dump should be in pretty much perfect condition.

comment by VPetukhov · 2019-12-14T10:24:19.408Z · score: 1 (1 votes) · EA(p) · GW(p)

Oh, sorry, didn't figure it out. Thanks for clarification!

Do you by chance know why the old wiki died?

comment by riceissa · 2019-12-09T22:28:02.029Z · score: 7 (5 votes) · EA(p) · GW(p)

wikieahuborg_w-20180412-history.xml contains the dump, which can be imported to a MediaWiki instance.

comment by saulius · 2019-12-10T00:06:09.542Z · score: 6 (4 votes) · EA(p) · GW(p)

Do you know approximately how many monthly visitors the old wiki had?

comment by VPetukhov · 2019-12-09T23:16:57.440Z · score: 2 (2 votes) · EA(p) · GW(p)

Thanks for the answer!

the information in that wiki is now lost to the wind

Woh, that's sad. Some thing for us to keep in mind for the future...

This may or may not fulfil your needs (and isn't _quite_ a wiki), but the EA Hub resources (resources.eahub.org), is a repository for EA links, and we hope to grow the number of resources available.

Indeed, suggestions would solve many problems! Still, the question is how to make it appealing for contribution. I really like button "Edit" on wikipedia, as it doesn't imply complex underlying revision processes and gives me around as much power as I want...

Can you please share your vision on the optimal knowledge management within EA? @EdoArad provided quite some ideas on how it could be organized, and among other I agree with the vision that if anywhere, the new knowledge base should be kept under EA Hub domain. So if you think it worth discussing, I'd be happy to engage.

comment by Ben_West · 2019-12-13T21:07:39.367Z · score: 5 (3 votes) · EA(p) · GW(p)

But with all my respect to wikipedia, I think that having a local wiki would allow to focus on more action-related topics instead of some general knowledge

I'm curious to hear more about your concerns with just using Wikipedia. I agree that there will be some topics which are outside the scope of Wikipedia, but it seems like many EA-relevant topics are within the scope of Wikipedia, and do not have very well established pages. For example: there is no page on longtermism, cause neutrality, or the INT framework. Even the page on effective altruism itself is pretty short.

My guess is that someone could pretty easily just go through old Forum posts and copy facts into Wikipedia. E.g. the section on invertebrate sentience is two sentences long, and I would bet that a huge chunk of recent Forum posts on invertebrate sentience could be justifiably included in that Wikipedia article.

In general I have a lot of nervousness about trying to re-create an existing successful product (NIH syndrome), and my guess is that Wikipedia will be more considered trustworthy, get more views, and generally be more influential than a local wiki.

comment by VPetukhov · 2019-12-14T10:46:45.842Z · score: 3 (2 votes) · EA(p) · GW(p)

There are only two main concerns. The first was explained in details [EA(p) · GW(p)] by @saulius, and I share his vision about motivation. Having something local would allow us to design award system in the way relevant for EA community and infrastructure.

The second is different scopes of relevance between local wiki and Wikipedia. Let's say, "List of annual EA events in Europe" would be relevant for EA community, but not for the others. I'd even expect that it could be harmful for the community to have such info on wikipedia. Moreover, searchability of a local wiki is much higher than for the global one. Though I think that this part can be addressed by having properly organized resource list, open for suggestions [EA(p) · GW(p)].

So, maybe it's better to focus on designing some mechanism to motivate people to contribute to the resources and the global wiki instead of creating the local one. Don't know what would be easier, and perhaps the answer depends very much on how EA Hub is going to organize suggestions.

EDIT: After some more considerations understood that I missed one more important point: knowledge distilling. At the moment, to find the latest ideas on some topic you need to go through dozens of the forum and blog posts, pages of individual organizations, facebook groups, slack channels, etc. Reducing time for such "research" would be very helpful. And I have doubts that it can be properly organized through resource lists, as the knowledge is very spread indeed. And again, because this information is relevant only within the community, I don't think wikipedia is the best place for it.

comment by JP Addison (jpaddison) · 2019-12-10T18:30:39.374Z · score: 4 (3 votes) · EA(p) · GW(p)

I'm interested in the answer to this question; I was just discussing it when discussing the EA Forum on a team retreat. I'm interested in the creation of reference documents in general. Something I was thinking about very brainstorm-y, no plans to implement in the near term) was to add a type of post to the EA Forum that allowed either direct editing or inline suggestions by not-the-author.

comment by EdoArad (edoarad) · 2019-12-10T20:41:18.461Z · score: 3 (2 votes) · EA(p) · GW(p)

Like a community wiki on stackexchange? Sounds valuable. (I think suggestions should be a default)

comment by JP Addison (jpaddison) · 2019-12-10T21:54:44.014Z · score: 1 (1 votes) · EA(p) · GW(p)

Like a community wiki on stackexchange?

What do you mean by that?

comment by EdoArad (edoarad) · 2019-12-11T04:24:43.231Z · score: 6 (4 votes) · EA(p) · GW(p)

Some questions on Stackoverflow or other sites in SE are marked as community wiki. This means that anyone (above a minimum reputation/Karma) can edit the question or the answers, that there is no "main author" anymore, but instead a mix of authors defined by percentage of contribution, no one gets reputation/Karma on anything.

I think that the loss of authorship is important so that anyone would feel comfortable editing the question/answers to make it a better source of up to date knowledge