# The polymath blog

## July 26, 2009

### Mock-up discussion thread

Filed under: discussion,mock-up — Terence Tao @ 8:06 pm

This is a mockup of what a discussion thread for a polymath project would look like; it would run alongside the research thread, but is focused on strategy, exposition, and discussion by casual participants, as opposed to the more cutting edge research conducted by more active participants in the research thread.

Please feel free to make suggestions in this thread as to how the format and organisation of these projects (or this blog) could be improved.

## 23 Comments »

1. Some tweaks to the blog format here: I’ve changed the names for the “thumbs up” and “thumbs down” icons for comment rating from the generic “Rate Up” and “Rate Down” to the more specific “Useful” and “Needs work”, as this is what we really want the ratings to reflect for the polymath projects.

Also, this is not visible yet, but I’ve broken up the comments into chunks of 50 each; once the 51st comment is reached, the post will display the comments from 51-100 by default, with a link to go back to the previous comments. This may hopefully solve the problem of having to start a new thread every 50 or 100 comments or so. One can see an example of this at the xkcd blog, e.g. at http://blag.xkcd.com/2009/06/18/security-breach/#comments .

Please feel free, by the way, to suggest any design enhancements to this blog (suggestions from CSS experts would be particularly useful).

Comment by Terence Tao — July 27, 2009 @ 3:03 pm

2. In case anyone is interested, there were a number of reasons why we chose this particular wordpress theme (which is called Rubric, by the way). Firstly, it had automatic comment numbering; secondly, it was fairly uncluttered; but thirdly, it had a particularly wide column width for comments. I think part of the problem with reading 100+ comment threads, as is the case with polymath projects, is that the narrow columns of most blog themes make such threads extremely long. Presumably with more CSS customisation one can improve things even further, but I am an amateur only in these issues (but would welcome suggestions).

Comment by Terence Tao — July 27, 2009 @ 4:17 pm

3. Terry, I’d like to better understand the rationale for comment rating. Will the ratings be reflected somehow in the presentation of the comments? E.g., they might affect the order in which comments are shown (for an example, see http://news.ycombinator.com/item?id=726544 ), or might be used to suppress comments (like Slashdot). Or are they simply intended to help guide readers in some way, e.g., by helping highlight unusually insightful comments?

Comment by Michael Nielsen — July 28, 2009 @ 1:58 pm

• Well, as far as I know wordpress has no facility yet to order comments by rating, and this would interfere with comment numbering, so the former reason is not really a factor right now. The idea arose from the feedback from the minipolymath project, where it was noted that once the research thread exceeded 100 comments or so, it became difficult to catch up, because one did not know in advance what the key insights were, and also which arguments later proved to be incomplete or otherwise faulty. The hope is by supporting deep threading (currently we allow 10 nested levels of comments), together with ratings to identify particularly useful posts (as well as posts that ended up not working as intended), this may assist casual readers in keeping up. (Simultaneously with all this, we plan to have some volunteers summarise the research thread from time to time in the discussion thread. We will have to run another polymath experiment to see whether this all works, of course.)

Comment by Terence Tao — July 28, 2009 @ 2:37 pm

• So far I don’t think the ratings are working very well. I’ve read through the developing thread and so far the application of thumbs seems arbitrary; for example, why did this comment from Gil get a thumbs down?

Additionally, we’ve already discussed the importance of allowing mistakes and errors; the original screed from Dr. Gowers was to allow half-formed ideas, but I fear if people can start to get rated down that we will get less experimentation.

Finally if there’s some sort of filter code to only view comments over a certain threshold, I fear it could make things more rather than less confusing, because often key insights have only made sense in the context of the (more minor) comments that preceded them.

Comment by Jason Dyer — July 30, 2009 @ 2:47 pm

• Jason – I’d like to wait until there’s a bit more data on the use of the comment ratings. It’s possible, for example, that they may turn out to be unexpectedly helpful when people summarize blog discussion on the wiki, as a way of quickly identifying crucial comments.

Comment by Michael Nielsen — July 31, 2009 @ 1:59 pm

4. My own feeling is that ratings should not be used to change the order of comments, or even to indicate whether you think that a comments was a good one, but rather to indicate to casual readers which comments they should read if they want to follow the essential points of a discussion without reading all the comments. For example, if somebody asks a subquestion that goes on to get discussed quite a lot, then the relevant comment should be related highly. And if someone else finds a counterexample to that question, then that too should be rated highly. But if someone proposes an approach that is not taken up, then that should not be rated highly (unless you think that it is in fact a very important comment and should not be ignored). I’m not sure I see any useful purpose for negative ratings, but maybe someone can suggest one.

Comment by gowers — July 28, 2009 @ 2:41 pm

• I changed the labels for the thumbs up and thumbs down from “Rate Up” and “Rate Down” to “Useful” and “Needs Work” (you’ll see this if you hover the mouse over these labels). I agree that negative ratings should be used sparingly, but the one place where I see they could be useful is to flag a post which contained an argument which overclaimed a bit and was subsequently found to have some gaps.

Comment by Terence Tao — July 28, 2009 @ 3:41 pm

• If someone posted a flawed proof, I think that we should vote “Needs work” to indicate that the proof doesn’t work. On the other hand, the proof might contain some important ideas and be “Useful”. Right now, it is not possible to vote both, only your last vote counts. And the only way to cancel a vote is to vote the opposite. Is it possible to change this?

Comment by Sune Kristian Jakobsen — July 28, 2009 @ 5:00 pm

• Perhaps in this case many people would up-vote a “reply” comment explaining the error in the earlier argument. Whether the earlier argument got a lot of up-votes or a lot of down-votes or both, I think it would indicate to the reader that something important may be in that comment.

Comment by Ryan O'Donnell — July 29, 2009 @ 12:47 am

• By the way, I put a draft policy for ratings at

https://polymathprojects.org/general-polymath-rules/

but it, like all other polymath rules, are of course open to debate.

Comment by Terence Tao — July 28, 2009 @ 4:28 pm

5. There has been a request to move the comment metadata (author, timestamp, and “Reply” feature) to the top of the comment rather than at the bottom. Presumably this can be done by editing the CSS – does anyone know the precise string of CSS code that would do the trick?

Comment by Terence Tao — July 28, 2009 @ 3:40 pm

• Just to say that I made the request — the reason I did so was that I found that with the current layout, comments that reply to other comments look as though they have been written by the person who is named just above. For example, if you aren’t thinking too hard, this comment will look as though it is written by Terry. I have found this minor confusion slightly irritating, and I think my irritation would accumulate if I was in the middle of an active Polymath project.

Comment by gowers — July 28, 2009 @ 3:57 pm

• an easier solution is to space out comments such that they’re more distinct – it also improves readability. just add to the stylesheet:

#commentlist li {
margin-top: 40px;
}

Comment by bliu — July 28, 2009 @ 6:19 pm

• Thanks for this. I added it to the CSS on a trial basis. It seems like an improvement to me.

(I’ve found that it doesn’t always seem to take effect, though. I assume that’s probably some type of caching issue, and will disappear.)

Edit: I see what’s going on. It’s displaying correctly in a kind of preview mode available to admins. But apparently we have to purchase the ability to modify the CSS, for general use. I’ll look into it.

Comment by Michael Nielsen — July 28, 2009 @ 6:37 pm

• I’ve enabled CSS modification. Looks like the spacing has taken effect.

Comment by Terence Tao — July 28, 2009 @ 6:57 pm

6. Omg I knew that! Rubric has automated commenting.

Comment by Anonymous — July 28, 2009 @ 4:04 pm

7. Another point. I think we should revisit a discussion that we had during Polymath1 about threading. In particular, the questions I would like to discuss are these: how much threading should we have, and what should it be used for?

Here are my views on the matter. I think we should have only very limited threading, and a reply to comment X should be given as a threaded reply only if it is a very local matter to comment X rather than a continuation of the general conversation.

To illustrate by an example, if comment X said,

I’ve just thought of a rather interesting conjecture. It seems to me that $x^n+y^n$ can never equal $z^n$ when $x$, $y$ and $z$ are positive integers.

then the comment

You obviously mean $n$ to be greater than $1$, and the conjecture is false for $n=2$. But for $n\geq 3$ it looks interesting.

would naturally be a threaded reply. However, the comment,

I’m assuming you require $n\geq 3$ here. If so, it occurs to me that if the conjecture is false, then one could create some elliptic curves with pretty strange properties. In fact, if all elliptic curves were modular, those properties might even be impossible. Could this be a fruitful line of attack?

then it would be a separate comment, even if there had been several comments in between. (If that was the case, then one would give the number of the comment one was reacting to.)

What I am describing is roughly the convention that we settled on with Polymath1, where I allowed depth-1 threading only. But there is room for disagreement here. Basically, what people think about this depends on how much they value they attach to the linearity of a blog, which makes it easy to see how the thoughts are developing and to keep track of which comments have been made recently, and how much to the ease of seeing later which comments were linked to which. I myself attach a high weight to the former, with the wiki providing a counterbalancing tree structure.

Comment by gowers — July 28, 2009 @ 4:06 pm

• With regards to threading, I have a few comments. I was a casual participant in miniPolymath1, and I can say that it became too much of a headache to follow everything. Allowing depth-10 nesting of comments will certainly help, but I still think it will be intimidating for a casual reader to jump in simply based on the number of comments he has to sort through.

I am not sure that this is even feasible on these blogs, but I know how I would like to see comments (as a casual reader). I imagine a bulletin board style layout. At the top are a few links which look like
“Taking John’s idea from comment 10 further (59 comments)”
“Taking Dan’s idea from comment 19 further (30 comments)”
“Continuing on the ideas of comments 11-15 (100 comments)”
Below these links is the original discussion of the problem. People could continue to post observations which are disjoint from all the above links, such as fresh ideas, new approaches, and observations which apply across the board.
Clicking on “Taking John’s idea from comment 10 further” would lead to another page with the same layout. Again at the top we read “Taking X’s idea from comment n further” and below these links we see the discussion, but this time the discussion is only concerned with John’s idea. You get the idea.

To achieve such organization, perhaps those who are working full time on the project (or whoever has agreed to occasionally summarize the findings) could create these links and then be the first to post on the newly created page with their summary.

I am no expert on manipulating blogs so I understand that this may not be as easy as I’ve made it sound. But this sort of layout would be most pleasing to the eye of the casual reader who may just want to work on the most popular attempts (gauged by the number of comments next to each link). And it also allows more organization in the comments since we get to see the relative depth of the observation based on how many clicks into the thread we are.

Comment by Scott — July 30, 2009 @ 9:34 pm

8. Do you think it would be worthwhile to create a WordPress “page” for each project giving the problem statement and some background information and links? Or do you want to keep that sort of information just on the Wiki?

Comment by thlog — July 29, 2009 @ 2:47 am

• Well, ideally the wiki and the blog would be integrated, so that one could mirror a wiki page on the blog or vice versa (or even more ideally, there would be a single page object somewhere that both the wiki and the blog could access or edit, essentially simultaneously). [This, incidentally, is one of the capabilities of Google Wave.] But until such time, I guess the best option is to have a master page (on the wiki) for this sort of thing, and links to that page on the blog. So I’ve set up a wiki page now for all of the active projects, and on the blogroll on the right of the blog there are links to each of them.

Comment by Terence Tao — August 4, 2009 @ 3:04 pm

9. , ?

Comment by Mirtesky — May 21, 2011 @ 6:08 pm

10. Заброс удочки Одним из главных условий результативной ловли является грамотный заброс удочки, ведь неумелое обращение со снастью может распугать всю рыбу Удилище Kosadaka Cinergy Tele Bolognese Удочка на карпа Карп, который водится почти во всех водоемах со слабым течением, отлично ловится весной и летом в солнечную

Comment by bingolinn — June 18, 2012 @ 12:54 am

Blog at WordPress.com.