Costs and Benefits of Delaying Setup by One Turn

What are the benefits of delaying your algo’s setup by one turn at the start of a match?

I can think of two scenarios in which delaying your setup could be beneficial.

1. Your algo is adaptive
If your algorithm is adaptive, ie. at least part of its strategy is influenced by the moves its opponent makes, then waiting until turn 2 to place down defensive units could be beneficial. Depending on how adaptive your algo is, you could base your setup design entirely on your opponent’s opening, which would require you to delay your setup at the start. This has the potential to be greatly effective, as ideally your setup would perfectly counter your opponent’s, although that is limited by how adaptive the algorithm actually is.

2. Your algo is static, but asymmetrical
This situation is less beneficial than if your algorithm is adaptive, but could still have its uses. At the end of the first turn, your algorithm could decide which half of your opponent’s field is weaker, less defensive, more defensive, denser, etc. and mirror its design appropriately. For example, an algo could determine which side has more destructors and choose to place its attacking lanes towards the opposite side. While this application of delaying setup would be less useful than a fully adaptive algo, it could provide a strong advantage in the early game. Additionally, its implementation would be easier, as you would only have to look at each half of the opponent’s field for one specific trait, such as density or defensiveness.

However, there are some downsides to delaying your setup by one turn.

1. A maximum of five damage could be dealt without consequence
Losing 1/6 of your total life on turn one could be disastrous, especially if your algo isn’t very defensive. This could be the deciding factor in long games, where the bonus delaying has in the early game isn’t as influential. This would also prevent algorithms that make no moves the first turn from becoming too widespread, as algos that immediately attack would have a large initial advantage over algos that delay their defense.

2. Implementation may be difficult
Implementing an effective method to adapt or react to your opponent’s setup may be difficult. This can be made slightly easier if your algorithm only looks for a specific aspect in the opponent’s setup, but this is mainly applicable to static asymmetrical algos, not adaptive algos.

3. Algorithms that delay their setup counter each other
If two algorithms that are facing each other both delay their setup, then the net result is equivalent to starting the game with slightly more cores and bits than normal. While this isn’t a large con, it could potentially lead to two algorithms playing a game of chicken, where they both wait for the other to make a move until some critical point is reached and one or both of them attack.

In my opinion, delaying your setup any longer than the first turn is outweighed by the costs of doing so, especially as the amount of damage you could take increases every turn. While there do exist drawbacks to delaying your setup by one turn, the potential for an early game advantage may outweigh them for algos that have an aggressive strategy. This strategy, however, becomes less effective the more widespread it is, as other algos can implement strategies such as attacking immediately or placing a setup that is meant to confuse or misdirect algos that try to adapt or react to their opponent’s setup.


Good analysis. My thoughts are the same, with one further insight.

The fact that you gain a bonus core for each life the opponent loses gives a large advantage to aggressive algorithms. A magical super algo that learns and adapts and whatever else you want will still lose if it straight-up lacks the cores it needs to implement this perfect build plan it’s come up with. In my opinion, the best opening move, even if you want an adaptive algo, is to spend your initial cores in a symmetrical pre-designed plan that your algo can use as a foundation for any further improvising, as well as always being aggressive to secure those precious bonus cores for building.


Indeed, +1 core for every damage dealt is a big deal.

Just to toss another idea into the mix, you could try using scramblers to avoid being scored on on turn one if you are setting up no defences. A spread of scramblers will usually deal with pings, and may even score themselves.