5

you have to play a game with the following rule : white can initially play two moves, then one move each side as usual, and a draw counts as a victory for black. Which side do you choose ? (say differently, do you think this variant favors white or black ?)

Same question with three initial moves for white. Maybe this is already solved?

Mathieu Dity
  • 159
  • 1

3 Answers3

3

Since I'm a 1.d4 player, my first two moves as White would be 1.d4 and 2.c4. After that position, Stockfish gives an evaluation of +0.59 if Black plays ...e6. I would definitely enjoy playing this as White since the threat of playing e4 at any time prevents the Nimzo and Dutch. Black best option is to settle for a QGD one tempo down by playing ...d5 on the next move, but this would be very comfortable for me. So I'd probably choose White, but it's not an easy decision.

That being said, it's not close to winning for White. An extra move just guarantees a pleasant edge.

With three initial moves, I would go for 1.e4, 2.d4, 3.Nf3. Stockfish gives a +0.99 evaluation, and it would be very tough for a human to defend as Black in a practical OTB game. This seems on the verge of winning for White, but still not a definite win.

Inertial Ignorance
  • 19,818
  • 23
  • 70
  • As a fellow 1.d4 player, I think it's more principled to go for 1.d4 2.e4. IMO the main point of 2.c4 is that it's the next best thing after black has prevented 2.e4 with ...d5 or ...Nf6. – RemcoGerlich Jul 02 '18 at 10:06
0

It's a common knowledge that White have about 55% probability of winning the game over Black, for the equal strength opponents. Having one extra move at the beginning will give maybe extra 5%, maybe less. And one more move will probably add another 5%, bringing the probability of winning (given everything else being equal) to about 63-65%.

Since most of my games are not drawn, I would definitely go for an extra move(s) at the beginning, but statistically speaking, for the equal opponents with high draw ratio, this would be not advisable, they lose a lot of half-points for the 10% more wins.

So, here it is, if the opponents have more than 20% draw ratio, this new strategy has no advantage. If the percentage of draws is smaller than 20% -- a few extra moves will definitely help.

After some deliberation...

The effect of having a few extra moves will also greatly depend on time control the games are played with. The most impact it will have on the shorter games, with less pronounced effect for 30+ minute games or longer, that tend to end up in draw more often.

lenik
  • 1,726
  • 9
  • 27
  • 4
    Notice that the numbers you state, especially the 5%, don't really mean anything (I don't know how you derived those "probabilities", 10%, 20%??). White winning 55% or so in a database (for players of a certain ELO strength) doesn't mean probability of winning the game. – gented Jun 25 '18 at 07:14
  • @gented do you have any foundation for that or care to elaborate? – lenik Jun 25 '18 at 07:17
  • What in particular do you wish me to elaborate? I asked how you derived those numbers and how you infer that they relate to probabilities. – gented Jun 25 '18 at 07:21
  • @gented let's imagine we have the same engine playing itself, and out if 100 games 55 are won by White (for simplicity, let's say, there are no draws), what exactly makes you say that this has nothing to do with the probability of winning? – lenik Jun 25 '18 at 07:25
  • 1
    Because in order for this to be a "true" probability you must repeat the experiment for every opening, every set of initial moves, every middle game position and so on and so forth. At the moment most of the experiments with computers are done by using mainly opening books and not all initial opening positions are played - not to mention that as definition in game theory there is no such a thing as "probability" of winning: White either wins (first move advantage) or it doesn't (no first move advantage) - which still has to be demonstrated. – gented Jun 25 '18 at 07:28
  • @gented so if we have a 6-sided die, we cannot say the probability of rolling 4 is 1/6, because we haven't tried to throw the die from every possible height at every possible angle with every possible linear / angular velocity, right? =) – lenik Jun 25 '18 at 07:33
  • 1
  • If you want to derive the P from empirical experiments (as limit of the frequency for N approaching infinity) then yes, you would have to repeat the experiment with all possible initial conditions. 2. However, in the case of a die, the space of events is deterministic and the P measure can be derived therefrom (which is not the case in chess, as we do not have a table book of all possible chess positions and their results).
    • – gented Jun 25 '18 at 08:20
  • 1
    @gented ok, let's say, my numbers are rough estimations based on incomplete information, which still might give us some kind of "back of the napkin" approximation for the "real" probability, can we agree on this? =) – lenik Jun 25 '18 at 08:53
  • Ignoring draw in 1-0-0 payoff matrix is quite brave model :) – hoacin Jun 25 '18 at 09:18
  • @hoacin in the ELO2000+ database with30k games I have handy, there are only 8% draws, so we are safe as long as it does not raise beyond 10% (see my original post =) – lenik Jun 25 '18 at 09:47
  • 1
    8% draws is probably some mixture of bullet and Armageddon. That's really not common knowledge that over 90% of games with equal players end in win of one side :D – hoacin Jun 25 '18 at 10:23
  • @hoacin I've talked about the influence of the time control in the original answer. – lenik Jun 25 '18 at 10:29