On the face of it, berserk seems fair. In exchange for a time disadvantage, you get an extra point if you win. The trouble starts when both players berserk. Then both players get extra rewards without either being at a disadvantage. That's unfair to the other arena participants. There should be more of a disadvantage like you also lose an extra point if you lose.
The other problem is that there's a "classical" arena going on right now, and the top players are all really playing blitz/bullet. They have to play fast if they want to retain their position because the guy with the highest wins per hour is going to win the arena. If you actually want a classical arena where people play with slower time settings, you have to cap games per hour to 2-3 games. If you cap it at 2-3 per hour the arena has been going, people that start late could potentially catch up by playing fast.
I don't know if anyone studied the mathematics of berserking but I would bet berserking is not as advantageous as you think. Unless you massively outrate your opponent it's not worth risking losing (or building) a streak.
We can just look at the current top 20.
1 SafeTheWales 97% berserk rate
2 Entkoter 50%
3 Soltan1800 100%
4 subrha2019 71%
5 jeffbutt847 100%
6 Superman89 92%
7 Fasmc17 97%
8 Farmville 86%
9 Bksinn22 56%
10 runtilmorrow 70%
11 Chessedition 17%
12 SuperSenya 92%
13 freehumanity 96%
14 Chaozu 92%
15 DrLaskerAdvance 55%
16 toosoob 100% 1725 rating with a 43% win rate
17 Fuchsi 33%
18 ap87 100% 1708 rating with a 31% win rate
19 sisalspikes 95%
20 Landau96 67%
3 out of the top 20 players berserk less than 51% of the time. The only 1700s on the scoreboard berserk 100% of the time. If avoiding berserking was a superior strategy, it'd be reflected. As is, "classical arena" is another berserk blitz arena.
Even if berserking didn't exist, blitzing out would still be a good strategy because you want to get as many games as possible in. Arena is probably just not an optimal format for long time controls but other tournament formats could have the issue of long waiting times (if you work with rounds for example) or could make it more difficult or impossible for players to late-join (elimination-style tournaments). That makes Arena a good format for online chess but is indeed not so great for Classical -- Lichess initially didn't have Classical Arenas, they have been added by popular demand.
Sure. Berserking and blitzing are different problems. My original suggestions address both.
If you cap the number of games people can play at 2 * hours the arena has been going, people would be able to catch up if they joined late by playing fast. It currently isn't possible to catch up with a good player who started at the beginning and blitzed straight through it. Lichess thinks classical games normally last "25 minutes or more," so if that's accurate, capping the games shouldn't leave much time "between rounds" (assuming berserk is also fixed to not be the optimal strategy). That said, breaks are a feature. Playing chess for 9 hours straight like SafeTheWales did to win is grueling.
Yes, and that is what makes Classical arenas so hard. Plus, if you cap the number of games you can play, it is basically a round robin but you can play your rounds whenever you want. Also, you can purposefully pause and unpause so you can play with lower rated players. You can't really do that with the current rules because you are trying to win as many games as possible.
We can actually simulate this (greater penalties) using the latest tournament results. We could then find out what the optimal penalty should be.
In any case, I don't fancy myself winning that tournament. I just use it as practice for our local otb rapid tournaments. So I play really slow.
Look at the titled arena, where almost no one berserks. I also doubt that it's as advantageous as some think.
The titled arenas I'm finding are all bullet. Bullet is a different beast. It's not surprising they don't think 30 seconds is enough time to play a game.
And in long time controls they berserk because it's too slow.
