[msheald]

Hello! Just curious if there is anything different about robot games compared to in-person games.

I'll play the free 8-hand games in the mornings. I usually do mediocre, but it is free and enjoyable for what I expect.

When I check scores, I usually see a number people with scores over 80% - even 90%! For ACBL club and tournament games, those scores would be reported in the ACBL Bulletin on the Congratulations page. The sheer number of scores would seem to be highly unusual.

I am curious - what is different about robot games? Are the deals different? (I know declarer usually gets the best hand). Finesses have been discussed multiple times in the forum. It just seems like a different game than in-person games. Enjoyable, but with different playing philosophy and expectations. Best regards,

Mike

[mycroft]

This is what happens when you play 8 boards. Each one is worth 12% of your score, and it's quite possible to get 10 of that on every board. For the best players, maybe one time in 20?

In one of the "club games", you then have to do that again for the next 8. And the next 8. And now that one in 20 is one in 8000 - probably worse than that, because there's three chances for the flatter in there where all roads lead to 4♠=, with maybe one pair missing game and one blackwooding and getting to 5.

And one in 8000 is, for a single person, once in 20 years if you play one session a day. And those are the best - the ones that expect to get 60% in a normal (24-27 board) game, and are a bit disappointed if it's not 63, 65.

[johnu]

As mycroft mentions, smaller numbers of boards in a tournament can result in higher winning percentages. A good player may play perfectly for 8 boards, and your opponents may never make a "good" play in 8 boards. Over a longer tournament that won't happen. And there's luck. Maybe the winner played a lot of low percentage contracts that all happened to make on a lucky lie of the cards and/or bad defense. In another set of 8 boards, maybe all those low percentage contracts go down and your opponent scores in the 20-30% range.

Another factor is the low quality of play from most of the "free" game participants. Many are beginners or novices and rarely achieve the par result. Others may be just passing time and aren't really paying attention because it is "free". An expert player might be expected to score 60-70% on average. If they get pretty lucky, 80+% is within range.

[pilowsky]

Another big difference is that you can make any crazy bid you want to deceive the robots because there are no directors and basically no rules.
Deceiving robots is part of the fun.

Except for the Zenith Daylong (and some types of challenge games), robot tournaments are always best-hand; as South, you will always have the most (or equal to) high card points.
Best-hand may be a bit of a stretch in some cases when the actual value of a robot hand may be a lot better than yours in terms of shape+HCP.

In this topic of the Forum, you will find many threads discussing ways to play against GIB that (probably) don't translate well into normal practice.

[msheald]

Thank you! I appreciate the guidance. Best regards.

Mike

[wbartley]

It's mostly that the level of play online (not just in free tournaments) is abysmal. This results in a greater standard deviation of results. Combine that with the few number of boards and you end up with what you see.

[2200]

The essential difference between bot bridge and "real bridge" is, bot does not play mind games.

It means, all the cards are "honest".

For example, if AKxxx vs xx，you cash an honor, and see 2 appears, then the break is 3-3 for certain. However, against human, it doesn't mean anything.
In the extreme case, even playing slams, bots follow honestly, so it helps a lot when you have 9 trumps missing the Q, and have to decide whether to play finesse or drop.
Also, say you decide to play strip squeeze, against human, especially experts, it all depends upon you table smell. Tough opponent will bare his K at early stage to give you a nasty guess.
Against bot, there is no such thing. You can 100% safely assume that, bot will guard its K.

Another example, say you have AJ1098 vs xxx, you need double finesse. Against bot, the first time you lose to K, then you can be 100% sure, the next finesse is on.

Let's say you are in 3NT, dummy has AQJ10xx and you have xx in hand, with no other entry. The first finesse hold, but against human, a champion player would rountinely duck with Kx doubleton. Against bot, there is no such thing.

It also helps a lot when you have to find a 2-way Q. AJxx vs K1098，just lead the J, if it doesn't cover, it doesn't have it.




In bidding, GIB has serious trouble value distributional hands, and it doesn't apply pressure.

For example, against human, auction such as （1H)-1S-(2H)-2S-(4H)-??
Often you have to guess, should I sac or not? Tricky opponents would bid game anyway, waiting to hammer you for phantom save.
Against bot, you don't have to worry "should I compete 3S？ What if I push opponents to 4H and find out unbeatable?". You can fully trust their judgement. If they don't bid 4H, they have balanced hand and enough losers.

In conslusion, psychology plays a big part in real bridge game, but bot bridge doens't have it, yet.