We could potentially expect more people with duplicate accounts, but this, as previously, could be avoided by running a background check in case there is a strong suspicion.
One might ask another person to create an account on his/her behalf. Again manual verification would be needed to avoid giving away couple of hundreds to already existing player.
Benefits
By doing this, there would be better chances for retaining the persons who could contribute to the whole community. Rather than scaring them away by giving them almost no chance of winning we could welcome them with sense of accomplishment.
Potentially we could see fresh strategies and learn from them. I of course do not have high expectations, however, it does not have to be the whole pattern. It might be something around the efficiency.
Easier for ambassadors to promote the game.
Neutral IMO.
We could see some of the static metas like baby sawtooths or baby transistors winning these competitions, but this means that people who won invested their time to research, participate reading forum posts and essentially made effort to get there. For those who have been along longer, would be a nice topic for conversation and a hook to get them continuing, thus contributing.
I think instead having competitions with an ELO limit would instead be better, for example there could be a contest for people with their best algo being between 1200 and 1500 after X amount of time on the leaderboard.
You could get the problem of people making algos that purposely only reach within a certain range as better there chances of winning prizes, but to combat this it could made so to enter the competition you must not have had an algo on the leaderboard that has reached or gone past the limit, that way the competition would only be for players of a certain level.
I think the top 10 players being the only ones with a chance to win is indeed a problem, but I am not sure whether this helps. Some other ideas to address the issue:
Modded tournament
A tournement where the rules change slightly, maybe with a temporary extra mechanic that sends everyone back to the drawing board.
More byob competitions
So beginners have more access to strong algos to observe and practice against
Guide competition
People try to make a guide for the game and the best few are published for a price.
Collaboration competition
A player sends a submission where he either independely controls when/where to spawn units or where to spawn firewalls. Then in the competition two different players team up to form an algorithm that way. (This idea might be difficult to pull of though)
What about a tiered system. The top 10-20 players are tier 1, they are eligible to enter the contests with the highest prize pools, 20-50 can have competitions for smaller prizes and 51+ can have smaller prizes still. The payouts would have to be structured such that it would more advantageous to be at the bottom of a higher tier than the top of a lower one. A prize structure like the one used for the season 1 finale competition should work.
Hmmm this idea would probably work well. You dont even need the bottom of the top 10 to be more lucrative than the winner of tier 2. You could have the 10th place get 50, and the winner of bottom bracket get 100 and no one would complain I think. This would only work for a large enough prizepool though.
What also might work is giving the top 10 a prize on their head for defeating them. Juat eliminating a good algo will net you a prize. Maybe the rop players already get a portion of the prize on their head or something.
I think it would be fine if the winner of the 2nd tier got a higher prize than the worst member of the tier above, but I think it would be an issue if the expected returns were such that people were sandbagging to get into the lower tier.
This I don’t like as much because of the non-transitive nature of this game. The top algos are at the top because of general consistency across lots of games, but not necessarily because of any particular wins or losses, and can still be beaten in a non-transitive way with focused iterative design on a “top-10-slayer”, that beats a few of the top 10 but has poor performance against most other random opponents, and this shouldn’t necessarily deserve a prize.
I like several of the ideas here, and want to mention a couple of things. I personally like some sort of tiered option as well. There are, though a couple of issues with it. The primary issue I see is for someone like me, who would be sandwiched between to brackets (or tiers). My algos do well enough (usually ranked between 20-30) that they would not win any bracket despite being better than those who would still be rewarded.
In my opinion, a tiered system works best not using score as a delimiter but instead using time joined as originally suggested. This gives the desired effect of giving new users a realistic chance, while still rewarding everyone equally if your algo is really better.
