Computer loses every game.
I would think that a computer would lose many games, but not every game unless the hustler has found a bad opening line the computer insists on repeating. Once in a while a hustler would hang a knight or a bishop. There are just so many computer type tactics that just don’t compute with human players.
My money would be on the engine.
Hustlers tend to rely on psychological components that would have zero effect on an engine and the engine would have no concept of playing at odds but merely maximize its opportunities with whatever material it was given.
400 Quatloos on the newcomers.
Knight odds folks! Anand could probably beat me majority of times at knight odds but IBM sure as heck could not.
I said NYC baby – Washington Square baby. Get real. Forget deus ex machina.
Man invented the machine not vice-versa. And chess hustlers can play baby! Ask many an outatown chessmaster who arrived in NYC in a cadi and left in a bus.
A while back, Andy Soltis did a column on material odds games. Of particular relevance to this conversation is the story he told about an odds tournament in NYC. (I am working from memory here, so I apologize in advance if certain details are incorrect.)
At the Marshall CC, there was once an (obviously unrated) odds tournament that had normal Swiss pairings, with the following exceptions.
The lower rated player always got white.
Material odds were granted based on rating differences. For example, if you were rated 500 points lower than your opponent, you started with an extra rook.
The results of the tournament surprised me. According to Soltis, the higher-rated player won every single game except one - in the last round, an underrated A player successfully accepted a rook from Arthur Bisguier.
Clearly, I don’t know if the same results would hold today. And, of course, I’m not a master strength player myself. However, I find it hard to believe that a 2700 or 2800 rated human would lose every game to even a 2200 or 2300 rated player spotting a knight.
Throw in an abbreviated time control, which favors brute calculation ability, and a computer’s imperviousness to tricks and pressure that could gull a human, and I would have to say it’s even less likely that the OP’s postulation is correct.
[EDIT: changed a word from “higher” to “lower”.]
Boyd:
If a rated expert like yourself given kniht odds couldn’t beat a GM every single game please consider tiddlywinks. At max the difference may be a tempo or two.
When the facts do not work in your favor, (a) change your premise, (b) throw in some insults, or (c) both. I will give you this - your argumentation has proven amusing, if utterly unconvincing.
A (presumably strong) computer was what you postulated originally. A computer, especially in a fast game (which is what a Washington Square regular would certainly be playing), has a big advantage, due to the game being more tactical by necessity.
Moreover, I do not know that I could beat even a human GM every time if spotted a knight. I have never tried it. Have you? If so, please share your results. If not, please at least admit that you are speculating, rather than speaking with some experiential basis.
I can at least say that I have seen enough GMs win hopeless-looking games against 2000-2300 players that I would not rule out their chances, as you are so quick to do. I also know that computers are in their element in tactical situations, which are going to arise in fast games.
I won’t dignify the rest of your “argument” with a meaningful response.
How was the bus ride back home?
Yes, I grew up there and learned to play chess in Washington Square.
My money still on the engine.
Uh, how many odds games are in a chess computer’s program to date? Zero.
How many odds games has a chess computer won? Zero.
Talk about MY speculation!
Untrue.
There have been several computer vs GM games in which the engine has given odds to the GM. Do some homework.
The engines have the plus score.
SEE: Ehlvest v Rybka
And 2006 Chess life article by Larry Kaufman.
“Mr. Wonka calling on line one…”
As Mr. DeCredico correctly noted, it is quite simple to look this stuff up. Go to Wikipedia and search for “human–computer chess matches”. There’s a whole section on the various odds matches that have been held with Rybka. It has scored reasonably well giving odds…to GMs…in classical time controls.
Sooooo…care to try again?
Wasn’t referring to pawn odds but knight odds.
But I liked the Wonka video. You do have a sense of humor. Thanks!
Deep Rybka running on a newer machine would beat Anand on even terms. She’s a full rating class higher rated.
This should be easy enough to test. Find somebody with a newer quad-core and one of the top engines, and play a few games at Knight odds. You could do that this afternoon and publish 'em here this evening.
Given your rating of 2024, I predict you’ll get creamed, but I’ve been wrong before.
A knight is a lot (read too much) to spot a rated expert by a computer. Even a class A player has a good chance. Otherwise ratings are completely overrated/inflated.
By the way I just beat my young granddaughter at knight odds— so there!!
How do you know this? It’s possible you’re right, but I very much doubt it.
At least one of the engines has already given a top GM pawn and move. I believe it was an eight game match where the computer played Black each game and removed a different pawn each game. The computer won decisively.
I’m unaware of any similar matches reported with low Experts getting knight odds, but, as I said earlier, this would be relatively easy to test. Since your initial post referenced hustlers, I’m assuming the games would be at a fairly fast time control, which would tend to favor the computer.
My computer, running Deep Rybka, is about two years away from being state of the art but it’s still pretty good, and my rating is roughly comparable to yours (1986 USCF, 2019 FIDE) and I just played a few three minute blitz games getting knight odds, the scores of which will remain my secret shame. 
I predict in an eight game knight odds match at G10 against engines such as Deep Rybka, Deep Fritz or Houdini running on a current high-end home PC, you’ll be lucky to score a single point.
In a time control of g/5 or faster, a 2000 would score 0% at piece odds.
If the time control was 40/2, or some other type of slow chess, a 2000 player would not score 0%, I think. Maybe 10%?
I am about 600 points worse than Houdini at blitz, and I am a favorite in blitz with piece odds against a 2000. I would expect to score about 30% in slow games, but, as I said… I aint Houdini.
Not sure what strength to give a NYC Chess Hustler, but others here seem to be using 2000. Depends on the “hustler.”
In 5 minute or less, I am an underdog to Houdini in 5 minute, at piece odds.
BPF
To: BPF
Thanks for the assessment of your own game vs. computers. Opened my eyes.
DWL
If the computer is set with a high contempt factor and avoids exchanges, the human expert, playing classical openings, will bail out into pawn-up endings quite a bit (sacking a piece for a pawn to break the machine’s initiative). Even here, I like the machine’s chances. (Maybe I once drew a blitz game against Houdini at knight odds, but if I did, I don’t recall it. I do recall a couple dozen losses. 
)
GM Finegold’s assessment is slightly pessimistic, I think, but a 2000 player scoring around 10% at time classical controls seems approximately right. The human would win once in awhile!
To billbrock:
If computers lose once in a while it just shows they are being human too.
It’s kind of a bummer to have played chess for 60 years, becoming an “Expert”, and suddenly realizing that I am just a patzer worthy of Queen odds given by a contemptuous computer program.
DWL1945