Robin Hanson's posts from the AI Foom debate are not included in the list of all articles. Covering only Yudkowsky's side of the debate would be a little strange for readers I think. Should we feature Hanson's posts (and others who participated in the debate) during that time as well?
Very good point. I will try to get some consensus on which posts will make it to the rerun listing. I have no objection in principle to including non-Eliezer posts.
Good point. Posts with a bad summary in the wiki are a lot like posts with no summary, except it's harder to see which ones they are.
We have a few options for dealing with them. What I hope will happen is that people will write better summaries and add them to the wiki in advance. That way the posts can go live with good summaries, plus we'll be improving the wiki.
Another option is to not include summaries in the posts. I think that some people will find the summaries useful (e.g., in deciding whether to go read the post), so I'd rather have summaries if we can.
Option 3 is to require the person who makes the post to write a summary. I'd prefer to make it as easy as possible to make each day's post, so that there's as little friction as possible in getting & keeping this thing going. I'd also like to avoid the situation where one person puts some thought & work into preparing the day's post, only to have someone else make the post first. That's why we made it a simple template to fill in.
A fourth option would be to have people write summaries after the post is made and leave them in the comments. Then the post could be edited to include the best summary from the comments. That could work in combination with one of the other options, but I think it's generally better to try to get a summary done in advance, even if we sometimes replace it with a better summary from the comments.
Any better options that I'm missing?
I'm willing to write summaries of the articles, since I seem to be good at that kind of thing. I'll start adding them alongside the links on pages like this. It will look a bit messy at first, but we can reformat the summaries that others and I add, and then perhaps add them to the top of the articles (with credit to the summarizer).
Thanks to jwhendy for pointing me to this unmet demand.
Edit: My first attempt at a summary.
Great! Your first summary looks good.
The wiki has post summaries in two places. There are pages for summaries by year (like this for 2007) and there are summaries on the sequence pages like the one you linked. Right now, it looks like the same summaries appear in both places. So your summaries (and anyone else's) should get added to both places on the wiki.
The summaries by year have every post in chronological order, which matches how we'll be doing the sequence reruns, so we'll probably be copy-pasting most of our summaries from there.
Giving credit to the summarizers could be tricky since the wiki doesn't identify who wrote what. I guess there's information about who made each edit available in the article history, but that would take some work to dig up. Is there a simpler way to do it?
One way is that, if you add a summary, sign it (with ~~~~). I went ahead and did that with the summary I added today. Others who think they can offer an improved summary can append and sign theirs.
The disadvantage is that it adds clutter and breaks the convention of keeping signatures on the discussion pages and off the main. And assigning individual credit for summaries isn't strictly necessary, but it's a great motivator.
Another thing that we could do is make a thread in the discussion section about this, which will give people someplace to post their summaries and receive plaudits. That could also attract more attention and encourage more people to get involved.
Alternative:
This takes place on a new subreddit created for this specific purpose. I'm in favor of this, but it requires some admin action to move forward rather than being strictly up to the users.
That would be great, but we can't depend on it. The good news is that it's a minor modification to the plan as discussed above. If anybody out there can get this done, I'd be happy to support using it. Otherwise, discussion should do just fine.
I'm pondering the trivial inconvenience of clicking through to the old post and then returning to the new post to comment and discuss, and wondering whether it might not be better to reproduce the whole text of the post in question each time. Perhaps that's a bad idea, but worthy of consideration?
One reason for not doing that is that we would like to encourage upvotes to the original articles themselves. Can this be solved with software? Yes. Will it? Unlikely.
I put the sentence about Eliezer getting the karma he deserves into our post, and I mostly meant it as a joke. It's true that the OB posts have gotten less karma because they were posted before the introduction of the karma system, but I don't think anyone actually feels sorry for Eliezer for not getting that karma or worries about the injustice of it.
The genuine advantage that I see to getting more votes on the old posts is to make it clearer which sequence posts people liked, which is useful for anyone who wants to pick out the best posts from the sequences. I also don't think it would be good for the new posts to get lots of upvotes based on the quality of the original post that they link to (or reproduce).
Yes, I didn't think of that. Why exactly do you want to encourage these upvotes? It's not as though EY is short of karma, so presumably for some reason relating to the posts themselves?
Actually, I misspoke. The reason is not mine, but a perception I've gathered from the community. I presume the reason is to restore some sort of feeling of justice, as the sequences have not been upvoted to the level that other posts have (being pre-karma themselves) even though they are more influential.
[rant]
As far as I am concerned, I think the sequences should have been exempt from the Karma system. It makes no sense to start a community and seed it with 2 years worth of posts which can be used to collect karma. It renders the karma system and the 'Top Contributors' list in particular meaningless. My scraper says that the sequences have gotten 7585 votes, accounting for 75850 karma points, more than the difference between Eliezer and Yvain (and that's without counting upvotes on sequences-era comments).
I don't expect others to agree with me, and having maximum consensus about the sequence reruns is more important to me than trying to rectify my pet peeves about lw, especially this one that will no doubt be an intra-lesswrong mindkiller. So I have just gone with the [perceived] majority on this.
[/rant]
If you would like to set up a poll to see what everyone really thinks about this, that'd be great.
Come to think of it, a reason that would get me to support linking to the original would be that this can help produce community-authored summaries for the discussion posts, which will be a lasting contribution of the sequence reruns. But posting whole-text and producing summaries may not be incompatible if done right.
I think the point about getting summaries written is a good one, provided they're of a high standard.
Personally I'd agree with your rant, but it doesn't seem to be a point of too much importance either way really.
I generally think it's bad form to duplicate content, and I'd be worried about confusing readers (especially ones who happened on the new post without being very familiar with Less Wrong) and messing up search results.
But I see your point about the inconvenience, especially with comments back on the new post, so I'd be open to doing it that way if other folks are in favor of it (and it gets the approval from Eliezer, whose posts we'd be reproducing).
I'm not convinced that duplicating content from the same site by linking to it is any better form than duplicating it by copying and pasting (assuming the author's permission, of course). Which is to say, assuming Eliezer's agreement and the general community consensus that reruns are cool, both seem to be about equally good form.
How do you two, and the rest of the community, feel about waiting a while for the [exercises] to get going and doing the two projects concurrently? A poll follows.
I would favor prioritizing writing exercises for recent or upcoming reposts, and including any exercises written so far in the reposts. But tying the two projects together explicitly is courting failure; a delay in one should not delay the other.
If a while is until, say, May 1st, I would favour holding off. But it needs to be something concrete, and measured in weeks, not months.
I'd prefer to get started, unless there is some clear and specific benefit to waiting which you can point to. These posts are going to come out gradually over the next 23 months, so there will be plenty of time to prepare most of the exercises. It looks like the earliest post for which an exercise is in-the-making is Making Beliefs Pay Rent (in Anticipated Experiences), which we're scheduled to get to in early June.
Coordination doesn't have to be perfect for the projects to benefit each other. If an exercise gets completed before the rerun post has been made then the post can still link to it, and if the exercise is completed a few weeks after the rerun post then it still gets the advantage of addressing a topic that's relatively fresh in people's minds.
Okay, it looks so far like opinion favors just doing the reposts and letting the exercises take care of themselves. That's fine by me. Will the reposts be in chronological order, and if not, can you tell me what the first five or so are? I'm taking this repost project as an excuse to pick up the pace on the exercises.
POLL: Should discussion take place on the original post (with people leaving comments to Eliezer's post) or on the new post (leaving comments to post in the discussion section which links to Eliezer's post)? Vote here:
original post
new post
karma balance
And here's somewhere to collect discussion of this question so people can find the poll options!
Strongly favour having discussion on the new post - I think "re-booting" the discussions is one of the biggest benefits of this proposal.
I disagreed, but I would reverse my vote if there were also a way for the old, original posts to link visibly to their corresponding new discussions. (Not just in a tacked-on comment at the bottom.)
POLL BACKGROUND: the majority of voters would like LW to have a new place to discuss each of the old posts in the Sequences.
If you are not in the habit of watching or subscribing to /r/discussion/ or /r/discussion/recentposts/, please do not participate in this poll.
POLL:
I watch or subscribe to /r/discussion/ or /r/discussion/recentposts/, and I do not mind if 702 notifications of new places to discuss the old posts appear there over the next 702 days. Vote here.
I watch or subscribe to /r/discussion/ (or /r/discussion/recentposts/) and I do mind if 702 notifications appear there over the next 702 days. Please find some other place to post these notifications. Vote here.
Coauthored by Alexandros and Unnamed
The suggestion to bring back the sequences in a format that is more similar to new blog posts was well received, so the two of us have been working to figure out how to make it happen, taking into account the feedback received on that thread. We've come up with a plan that is ready to be put into action right away, and are presenting it here for further feedback before getting started. Let us know what you think – both whether it's worth doing, and what specific changes could be made to improve on what we have so far.
The plan is to have a regular “Rerunning the Sequences” feature in the discussion section, with one post each day linking to one of Eliezer's old posts. We will go through all of Eliezer's posts in order, minus open threads, administrivia, and quotes threads, following the list here. Starting with "The Martial Art of Rationality" and finishing with "Practical Advice Backed By Deep Theories", we count 702 qualifying posts (with help from Alexandros's scraper), so almost two years' worth of continuous posting.
The new post isn't meant to contain original content, so it will follow a standard template (a draft of which is included at the bottom of this post). The template includes a one paragraph summary of the article (extracted from the wiki), a brief explanation of how it's part of the sequence reruns, relevant links, and a standardized format for the title and tags.
Like the rationality quotes threads, this is designed to be implemented by the community rather than by software. It will only work if people are interested and participating. Someone will need to make the new post each day, and a lot of the time it will need to be someone other than us two. We'll post an html version of the template with instructions to make it easy for anyone to make a post with just copy, paste, and a few quick edits. The other way that we'll need folks to contribute, besides making the posts, is by writing summaries of the posts. The LW wiki already has a lot of post summaries, including the first 28 posts, but many posts still lack summaries. Things will work more smoothly if those summaries get written in advance and added to the wiki before we get to them.
The main purpose of the Rerunning the Sequences series is to make it easier for people to read or re-read the sequences by putting them in a convenient format that's more like reading new blog posts. Hopefully, this project will also be complementary to other sequence-related projects. As we go through the old sequence posts, some people may be inspired to write condensed versions of them, create exercises for them, or even just to add better summaries to the wiki. And perhaps Eliezer will finally be able to get the karma he deserves for his pre-LW posts.
Another advantage of the sequence reruns series is that it creates a focal point for discussion of the material in the sequences. It's much easier to have discussions about old posts when a lot of people are reading the same post at the same time. We've gone back and forth on where this discussion should take place: on the original post or on the new post. The comments will be about the content of the original post, so it sort of makes sense for them to go there, and that will help keep all of the discussion about that post in one place. On the other hand, this is a fresh discussion taking place a few years after the original posts, and commenting on the new post would keep everything in the discussion section so that the main page “recent comments” feed doesn't get flooded with discussion of old posts. Right now the template is written to have comments go on the new post rather than the original post, but this could go either way. There's a poll on this in the comments; we should decide one way or the other, since haphazardly splitting the discussion between the two places is the worst option.
So, what do you think? Does it make sense to put these new posts in the discussion section? To do the whole Yudkowsky oeuvre rather than specific sequences? To have one post per day? Do you have a better name than “Rerunning The Sequences” (tag: sequence_reruns, title: [SEQ RERUN])? Does the post template look okay? Try to make comments actionable (polls are recommended) so that we can make changes over the next few days and then get started (assuming that there aren't major objections to the whole project).
Unless there are major objections, we'll try to get everything ironed out and start with the first post next Monday (4/18).
Template
Title: [SEQ RERUN] Why truth? And...
Today's post, Why truth? And... was originally published on November 26, 2006. A summary (taken from the LW wiki):
Discuss here. [???]
This post is part of the Rerunning the Sequences series, where we'll be going through Eliezer Yudkowsky's old posts in order so that people who are interested can (re-)read and discuss them. The previous post was The Martial Art of Rationality, and you can use the sequence_reruns tag to follow the rest of the series.
Sequence reruns are a community-driven effort. You can participate by re-reading the sequence post, discussing it here, posting the next day's sequence reruns post, or summarizing forthcoming articles on the wiki. See here for more details.
Tags: sequence_reruns