### djm03178's blog

By djm03178, 2 weeks ago,

This is a half-serious blog, so I don't have a strong opinion on this topic and I welcome others' thoughts.

For years I've been seeing people calling some contests 'speedforces', but it seems like people have different definitions of it. The most common situation where I see people commenting that a contest was a speedforces is when the contest was in general too easy and therefore they had to solve many problems in a short time. However, I have a little different opinion on it.

Speed is, in almost all cases, very important on CP contests, but there are cases when it becomes even more significant compared to being able to solve harder problems, and I think it's not just when the problems in general are easy; instead, I call a CF round speedforces for people with specific range of rating when most of them could solve $K$ problems but not $K+1$ and this range is very wide. So, with my definition this term is not universal for all participants, but instead it applies only for certain participants.

For example, yesterday's contest Codeforces Round 960 (Div. 2) can be called speedforces (no offense, I think the contest was cool) for people with rating 1800 or above because solving A-D led to performance of from 1750 to 2400, depending on how fast they solved them. The fifth problem was quite hard so having a little better skill to solve harder problems didn't help much for most people in this range, unless they absolutely outperformed as a Div. 2 participant. The range of solving A-C on the other hand, has only a range of from 1450 to 1750 performance, so participants in this range experienced less speedforces and being able to solve just a bit harder problem was more helpful than just solving fast.

With this definition, one of the things that makes speedforces is when the hardest problem is too hard so barely anyone solves it, because it means we need to divide the whole performance range into $N$ segments instead of $N+1$ (including 0 solve), so in average each range gets widened and more people will get stuck at each number of solves.

So unlike the other definition I assumed above, having easy problems doesn't necessarily make speedforces. It's more like when the first $K$ problems are easy to everyone but the $K+1$-st problem is relatively much harder so most people cannot solve it. If these ranges are even then it will still evenly distribute participants into each segment depending on whether they solved one more harder problem rather than solving easy problems faster.

I just wanted to bring this up because I've seen some people regarding it as a mere complain from having skill issues and not something that exists, but I want to say it's a natural phenomenon that came from difficulty curve of a contest and therefore it is a valid argument (whether it's an actual issue or not is another story).

• +144

 » 2 weeks ago, # |   0 orz
 » 2 weeks ago, # |   +3 I think you explained it very well. Indeed, a contest is called speedforces when the majority of participants face a large difficulty jump (maybe C or D). For example: this contest may be regarded as speedforces by those participants who are $\leq 1800$, since there is a huge gap in difficulty between B and C. When this jump is at harder problems, people don't call it speedforces since most of the participants solved as much as they could.
 » 2 weeks ago, # |   +16 Disagree with this definition; in particular, I think you absolutely can get a speedforces effect from just the problems being too easy.For the sake of an extreme example, say you have a contest in which there are 1000 problems, all div2A difficulty. You will not have any huge gaps between problems, in fact the leaderboard might look very reasonable as people's speeds are naturally distributed in some bell curve-like shape. But you cannot say that in such a contest, it is not the case that "speed is more significant compared to being able to solve harder problems".In fact, I'd argue a contest like this is in a lot of ways worse than your example, because speed is the only thing you can improve to get a better rank (while in your definition, you can still improve at problem solving to break out of the speedforces range).Now of course real contests aren't that extreme, but even then the effect of running out of time before encountering problems you need to think to solve is relatively common in Codeforces rounds (or at least it used to be 2-3 years ago; these days problems are thankfully generally harder. Except for cases like Codeforces Round 959 sponsored by NEAR (Div. 1 + Div. 2) where F and G were really easy for their positions).