×

# How do ratings work?

This discussion has been deleted!

Note by Finn Hulse
2 years, 10 months ago

Sort by:

Without going into technical details, we use a modification of the Elo rating system used in chess. Everything is given a rating value, and an uncertainty value. The amount that your rating changes depends on the relative rating value, and the uncertainty values.

When the uncertainty value is low enough, the rating is displayed. There is no hard and fast rule about the number of attempts / correct attempts. e.g. if 20 level 1's fail to solve a level 5 problem, we would not have much certainty about the rating. Same applies if 20 level 5's solve a level 1 problem.

As a significant proportion of submitted problems have issues with them, we have decided to take a conservative approach to ratings. As the quality of submissions improve (say due to having reputable problem creators, active reporting and responding, or member moderation), we can afford to be more lax. Until them, I rather err on the side of caution. Staff · 2 years, 10 months ago

Hmm. So can you yourself official-ize ratings at your will? That was another big question I had. · 2 years, 10 months ago

The answer is obviously yes. Do we do so? Not often. Instead, we may just set an initial guess-timate, and let the system take over. This is similar to when a member sets the initial level of the problem.

We will not entertain requests of "please official-ize my problem ratings," Staff · 2 years, 10 months ago

I see. Also, have you seen Plugging Roots into Functions? My personal favorite and only Level 5 problem I have. :D · 2 years, 10 months ago

Sorry, it went to level 3. · 2 years, 10 months ago

I know... *tear ;| · 2 years, 10 months ago

Haha... XD · 2 years, 9 months ago

Comment deleted May 14, 2014

Mmm. Good suggestion. :D · 2 years, 10 months ago

I'm also wondering the above points, although I don't seem to have been victim to any of those problems Finn mentioned. · 2 years, 10 months ago

Really? Here's an example of what I mean. If one of my problems doesn't take root immediately, it will be lost forever. :O · 2 years, 10 months ago

That problem wasn't very good anyways. Not creative at all; in fact, such a quadrilateral can't exist. It deserved to not be known, I guess. · 2 years, 10 months ago

Haha, yeah. Another example of very GOOD problems would be your Inequality 1, 2, and 3 problem set. Remember that? · 2 years, 10 months ago

Oh yea, I think I better add some more problems to that before sharing. · 2 years, 10 months ago

Yeah. :D · 2 years, 10 months ago

This problem's with me too! · 2 years, 10 months ago

@Finn Your examples do not support your argument. If anything, they demonstrate that things are working as they should.

The first (geometry) problem is wrong, and had it's rating removed, hence the "This problem does not have a correct answer. Do you know why?". For the inequality problems, note that the set has not been shared. In fact, these problems each been viewed by $$< 10$$ people, and attempted by $$\leq 1$$ person each. As such, I would be highly dubious of the system if any of these problems have an "official rating".

Edit: I have deleted Daniel's question. Staff · 2 years, 10 months ago

That's all well and good. :D · 2 years, 10 months ago

That's with me too! · 2 years, 10 months ago

That's with me too · 2 years, 10 months ago

@Suyeon Khim @Calvin Lin @David Mattingly @Arron Kau And all other staff members. :D · 2 years, 10 months ago

Oh BTW @Daniel Liu I'll give you a high-five over the internet if you can solve Plugging Roots into Functions it's my first truly Level 5 problem. Good luck! :D · 2 years, 10 months ago

I keep on getting the same answer over and over again, and your problem says that it is wrong. I submitted a dispute. I just can't see what I am not getting... plus, there is no solution submitted so far. Mind submitting a solution so I can see how my answer is wrong?

EDIT: I checked it on Wolfram Alpha now, and it also confirms my answer. · 2 years, 10 months ago

Comment deleted May 14, 2014

Are you sure? I used wolfram alpha to bash it after I gave up, and it confirms my answer...

Can you post a solution? Thanks. · 2 years, 10 months ago

@Calvin Lin Doesn't Newton's Sums work? I keep on getting the answer of $$\text{[REDACTED]}$$ when I use it; what am I missing? · 2 years, 10 months ago

@Daniel Liu I did not work through this problem, and merely trusted that Finn dealt with the reports accordingly. I have just reviewed it, and corrected the problem accordingly. Sorry for the inconvenience. Staff · 2 years, 10 months ago

Yeah sorry. D: · 2 years, 10 months ago