Counting upvotes/downvotes

by calcsam1 min read7th Aug 201119 comments


Site Meta
Personal Blog

I've recently posted several articles in my "Building Rationalist Communities" series. Some, like "Holy Books (Or Rationalist Sequences) Don't Implement Themselves" have been fairly popular, as measured by upvotes; others, like Community Roles less so.

But here is a small problem: I can see my net upvotes versus downvotes, but I can't see the numbers of each. Normally this is probably not much of a problem. But this series has been relatively controversial, and I would like to personally distinguish disagreement (high upvotes and high downvotes), from disinterest (low upvotes and low downvotes).

At least for my personal satisfaction and curiosity if nothing else, is there any way to see the absolute numbers? 

19 comments, sorted by Highlighting new comments since Today at 9:12 PM
New Comment

I don't see why this isn't implemented for both comments and posts. I always wonder when I go up/down if it was because someone removed a down vote or added an up vote or vice versa.

Provided we do not have access to the actual numbers, we can still attempt to build a decent map for the territory. After all, that's what LW is all about, right? So here is my first crack at it.

I'd guess that there is a correlation between the total number of votes and the total number of comments. My rationale: roughly the same percentage of people who vote also comment, regardless of the post.

If true, this can be calibrated using some popular but non-controversial posts, where nearly all votes are expected to be upvotes. Thus, in absence of better indicators, you can count the (top-level) comments, scale it by the (yet to be determined) comment/vote factor and get an estimate for the total number of votes.

Now, in the spirit of rationality, here is how this model can be tested: find several uniformly well received posts from different (popular) authors with a fair number of comments, apply the above metric (post karma/# comments) and see if it is reasonably stable (the value is within, say, 50%). A quick glance at some of the more popular posts suggests the value of .5 to 1.5 for karma/#comments, not accounted for controversy. If this holds, then having a post karma/#comments significantly below, say, 1/3 would mean that people downvote the post a lot, rather than ignore it. I suppose this is a long-winded way to say "just count the comments!"

This is, of course, only a zero-approximation map, and it's easy to see how one can improve it, but one always needs a starting point.

This feature has been suggested many times but isn't implemented. (Yet?) I'm not sure if it's waiting for code to be written or for grand high muckety-mucks to approve it.

Controversy scores would indeed be useful things - e.g., are the scores on the QM sequence so low because the posts are controversial, or because few people read them?

[-][anonymous]10y 11

Note that it wouldn't be that few people have read them, it's that few people post-OB have read them and voted on them.

Also, the material is fairly impenetrable. My model of the typical LW-er says they would prefer not voting down EY when he writes something opaque, but they wouldn't necessarily vote up either.

I'm simply not smart enough to understand the math. It feels dishonest to vote up something I didn't fully understand, but it also feels unfair to downvote something that would probably be right if I could understand it. I bet a lot of people didn't vote at all on the QM sequence, not because they didn't read them but because they didn't feel informed enough to make a judgement.

I'm simply not smart enough to understand the math.

You don't have enough mathematical maturity to understand the math. You are smart enough to learn to understand it, quickly.

If I may ask, at what point during your attempted reading of the sequence did you arrive at the conclusion that you couldn't understand the math?

The first few posts, with configuration spaces and complex numbers, didn't make sense to me even after reading them a couple times and trying to draw them out myself. After a while I decided to just sign up for physics in school, since I learn better that way.

I didn't have any trouble with the posts on Many-Worlds and simplicity, or with the philosophy/QM stuff.

I thought the math was utterly absurd until I read this and went "Oh, THAT'S what a complex number is? Gee, no wonder they're used in this!"

I got it off the comment threads on one of the QM sequences, and apparently it helped a lot of people there. You seem to have already found your own solution, but I figure it doesn't hurt to at least mention something that worked for a lot of others :)

After a while I decided to just sign up for physics in school, since I learn better that way.

Don't let the curriculum hold you back. Get a book, read it yourself.

Downvoted for unsolicited other-optimizing, especially given that it's reasonable to guess from the context that KPier learns better from traditional classes than from books, if e learns better from traditional classes than from blog posts.

For a smart student, standard curriculum is usually too slow (and too limited) for any subject they actually care to learn. Status quo can hold them back, since everyone else is just following the curriculum.

That learning from books is less optimal is irrelevant to the extent there is no other choice (but there are probably video courses to be found on many subjects).

In most areas of study, I learn well on my own/from books. For math, I don't. I thought this meant I was stupid until I read "Beware of Other-Optimizing" - now I just think it's something about the way my brain processes math.

Standard curriculum at my school, even in the honors classes, is fairly slow; that's why I'm taking three science classes plus math next year, which (from past experience) should be enough to keep me from getting bored.

That said, I have also bought some physics books to read on my own once I finish Godel, Escher, Bach, which is taking me a while. Do you have any specific suggestions?

Try Lawvere's "Conceptual mathematics" (it's written for advanced high school level). It'll demonstrate some ways in which a lot of math is very unlike high school math. (But do look for other suggestions.)

A little UI idea to avoid number clutter: represent the controversy score by having the green oval be darker (or lighter) green the more controversial the post is.

A shadow of controversy scores are available for comments by sorting by controversial. Articles sorted by controversy are here. There used to be a link to this page before the recent changes. Sorting by "hot" is broken, so I am nervous about the abandoned controversial list as well. ETA: indeed, it is broken and the "next" button doesn't really work. Eventually it cycles.

Sorting by controversy isn't very useful, compared to controversy scores, let alone raw scores. Sorting comments has some value, but nested comments hide most of the information. There are too many articles for a sorted list to be of value. In particular, it is not easy to answer your question, of assessing a particular post. (ETA: there is the option to sort by controversy a shorter list, such as the last month's articles. This may be of value.)

[-][anonymous]10y 0
[This comment is no longer endorsed by its author]Reply