Thank you for sharing your opinion on my last blog.
I considered what kind of participants I'd like to honor the most. It was those who have strong confidence and eagerness to take part in contests, not those who care about their rating too much or think like, "what if I can't solve a problem...".
Thus I finally decided to adopt an "open-then-rated" system.
Now it's time to determine the details of the new system. What I'm currently considering is the followings:
- If you try to open a problem in the first five or ten minutes of a contest, you see a Read Problems button, and pressing it confirms your rated participation.
- If you are late (or don't press the button in the first five or ten minutes), you are unrated, but still can read and submit to the problems. Also, your name will be shown in the standings and it counts for GP30 scores.
This is a draft, so if you have a concern or an idea, please share it here.
do I need to press "read" button for every question or pressing it once will count for all problem?
The button is for this, so you only need to press it once.
Will we become unable to participate as rated if we are late?
When are you planning to apply this new system? In particular, is it before next AGC(3/6)?
we may feel comfortable and easy to adapt if there is something like a testing round before the start of the new system.
I'm not an engineer so I don't know how long it takes to implement this change, but I assume at least in the next AGC we have the usual system. I also assume there will be a testing round or something.
Is this system to be applied to all rated contests (including ABC), only ARC/AGC, or only AGC?
For all contests.
I want an unrated "Read Problems button" to enter contests without waiting five or ten minutes.
I really hope that AtCoder's rating system will not change anything. However, I also understand your plan is a necessary change for AtCoder to grow into a big service.
there has been some instances where i have started contests after even 20-25mins, so considering i wont be the only with this kind of situation, i think the ideas of your prev blog post were better, if a vote is considered from me.
"Late-then-unrated" system seems good because cheaters can't gain advantages using their sub-accounts.
It is a great idea.
Some of my friends don't submit unless they solve all the 6 problems to keep their rating.
Can you submit the problem before you pressed the button? Should'n there be another button to accept your non-rating participation ?
The second part makes the first part redundant.
If someone pressed the button and didn't submit anything then he will be counted as a non-participant or a participant didn't solve anything and lose rating ?
Lose rating, that's the whole point.
Ah, interesting
Will you make it possible to tell if a contestant is rated or not from standings? This information is necessary for the difficulty estimation in AtCoder Problems (https://kenkoooo.com/atcoder).
We'll do that.
What's GP30?
https://atcoder.jp/posts/171
It is simple to workarround that button using an alt account. With literally only two more clicks one can read the problems without being rated. So in first place this offers another level of tactical behaviour in a contest which has only little in common with the spirit of competitive programming.
From my point of view this obfuscates the graphical user interface instead of adressing the problem.
When you increase the cost of something you get less of it. This is what the proposed rule would do.
The fact that proposed rule does not solve all problems is not a good argument. The litmus test ought to be whether the effects are a net positive.
Yes, you are right with this, increasing the costs of "bad bahaviour" is per se a good thing.
Here, however, the bad behavior is even rewarded. If you configure your browser so that you can switch accounts with just a few clicks, you will get a tactical advantage in return. I am sure we agree that this is not a good thing.
Even if you can read problems, it's pretty hard to predict your performance in 10 minutes. Even if you get that slight tactical advantage, there will be risk of forgetting to open problems with original account and missing a favorable contest. Overall a lot less people will be doing it compared to current number of people who attempt "submit only if performance is good"
Ten minutes is too small for a contest with 270 minuhtes duration (e.g. AGC 051). I think a better time limit for opening problems would be 1/10 of contest duration.
27 minutes seems too long though...
Why does the contestant has to press the button(i.e. decide whether rated) in 10min after the contest starts instead of 10min before the contest starts? Theoretically, the contestant should not be able to see anything before pressing the button, so this should be the same except that if it is after the contest starts, the contestants gaining knowledge of the problems (using alts, etc) before pressing the button gains advantage illegally.
It can often happen that some contestants get ready for a contest a few minutes after the contest starts, and I don't want to make them unrated for only a few minutes. In addition, I don't think 10 mins make a big differnce.
Maybe you can make a button called "I want to be unrated" and participant will be able to see the problems as well.
I think even if a user is late, he should be given the rights to have this contest rated.
It is great,and I won't care about my ratings.
What is the point of "Read Problems" button v/s just a "Join as Rated" button? The first one is still a bit unfair since people might use alts to read the later problems and in 5-10 minutes join if they're doable. In the latter, basically everyone can read any problem and decide if they want it to be rated in the first few minutes. So the latter is a fair system that makes alts totally redundant.
damn, I thought this rule was already implemented and started the contest 10 minutes after :(