Tricki
a repository of mathematical know-how

Rating articles

Rating articles

This is not so much a feature request as a feature proposal. Olof recently suggested to me that we might like to have a system for rating articles. I think there are advantages and disadvantages of doing this, and have been trying to think of a system that would maximize the advantages and minimize the disadvantages. Here are my thoughts so far – further input would be welcomed.

1. I am not too keen on making it possible to rate articles negatively. At this stage, we are trying to encourage contributions and get the Tricki to grow to a point where it becomes noticeably more comprehensive. So I wondered about having a button that one could press, which would mean simply, "I highly recommend this article". Articles that did not get lots of such recommendations would not stand out, since several pages are likely to be fairly neutral (such as front pages without much text, or very standard but useful accounts of certain techniques).

2. We would need to come up with some policy about what to do with the ratings when an article was substantially edited. One crude possibility would be for ratings to expire after a certain time (and be renewable by the person who made them). Other possibilities I can think of have serious problems with them. Perhaps it should not be possible to rate an article until it has been marked as sufficiently complete. Any thoughts?

3. In addition to a "quality rating" like this, we could have a "comprehensibility rating". The idea would be to have a scale such as 0=nonmathematician, 1=advanced high school student, 2=beginning undergraduate, 3=advanced undergraduate, 4=advanced graduate, 5=expert in specialized area. By making such a rating, one would be giving one's assessment of the actual requirements to read the article (and NOT – this is the point – what the author claims is the required background). This could provide useful feedback: if you intend an article to be comprehensible by beginning undergraduates but it gets rated as an advanced graduate article by several people, then you may wish to make changes.

This could be a nice feature. Comments:

1. In addition to the "highly recommended" button I would put a "stub" or "needs more work" button to alert people kin to elaborate alredy existent articles.

2. Seems to me that the exact tool to implement this are "social bookmarking tools for news", like those implemented on http://www.digg.com or http://meneame.net (spanish version), where users vote in order to get news up in the news stack. I mention the spanish version because:
a) I believe it has more and better features (http://es.wikipedia.org/wiki/Meneame#Funcionamiento)
b) It is open source, so maybe some material can be imported near-directly to the Tricki by Olof & cia.

3. The "articles stack" could replace the "list of all articles" page. So, if you wanted to look at the most recommended ones, you will find them ordered there quite like in Digg or Meneame (you should be able to "vote" for them there too).

4. There should be another stack for stubs/articles which need more work, something like "Needs Work stack". We could even have yet another one, the "Working now stack", where "live" articles that are being changed "now" (i.e., there's people in charge of develop them further). Of course, the three stacks don't need to be disjoint.

5. About editing: one could have a button to push when editing, which said: "Was that a minor editing or a major editing?" If it was major, its share of votes would start to decrease at some given rate, and an automatic message would be sent to anyone who had voted for it before, alerting them of this fact. This way, they could go to reread the article and revote it if they still found it great (btw, if they didn't find it as great as before, shouldn't them have a button to say "please, leave this great article as it was before"? Not for all articles, just for the "great" ones). New votes wouldn't get erased by the decreasing algorithm.

6. We must also be sure that no one votes an article more than once (unless it suffers a major modification). We should control voting IP's or let only vote to registered users. "IP control" is less reliable because of dynamic IPs, but would be the best choice if we allowed non-registered users to vote. "Registered users only" is more reliable and would make easier to alert them when a major change has been done to a specific article.