102

Data is considered, basically, a sentient supercomputer. The ultimate AI. He was built with an ultimate storage capacity of eight hundred quadrillion bits and a total linear computational speed rated at sixty trillion operations per second.

Chess is not a game of intuition or empathy, it's a game of computational skill/power and positional manipulation. Even today's crude computers can beat the best chess players on the planet.

Data is programmed with "extremely advanced" chess routines yet...he loses a chess game to a relative novice chess player. If Diana was a strong player who actually studied the game, she would be reasonably comparable to a modern day 1700-1800 rated 'B' level player. Data on the other hand with his 'extremely advanced' chess routine programming would likely be equivalent to a modern 2800+ rated Grandmaster. At the beginning of TNG 5x14 - 'Conundrum', Counselor Troi--having shown no particular skill much less highly advanced chess skills in her entire existence-- beats Data in a game of chess.

How can this be?

Improve this question
3
  • 8
    I made Data to be as human as possible! – Often Right Sep 7 '15 at 4:15
  • I think in general chess has been mistreated in movies and TV; heck, about half the time, the board is not even set up correctly with white's king rook on a dark instead of the correct light square. Star Trek once showed "superhuman intelligence" of alien entities by having them play chess (in the bodies of crewmen) and one making a move and the other instantly resigning implying that they knew the game was "solved" and so why would they bother playing -- go would have been a more plausible game or maybe even poker. – releseabe Feb 7 '20 at 0:59
  • 1
    Have a 100-score gold badge on me :-) It's my 10,000th question upvote on this site! – Rand al'Thor♦ Aug 9 '20 at 18:03

13 Answers 13

Active Oldest Votes
95

I wouldn't overanalyse this too much - it's just a case of bad writing. Furthermore, it was written in the early 90s, back before computers surpassed humans at chess, and people still thought that magical human intuition could beat brute force calculation every time.

Really, the ludicrous part is the idea that a 'classic attack', which has apparently been sufficiently well analysed that it has both a name and a 'characteristic response' which also has a name, could be refuted by a 7-move forced checkmate (over the board by an amateur, no less). This is typical ignorant TV chess writing, where the response to everything is an overlooked checkmate.

The whole scene is the chess equivalent of technobabble.

Improve this answer
5
  • 9
    It's surprising to me that there aren't more 'real' chess players here. As a rated Class A player, I find this to be the most likely correct answer and yet it has received only 1 vote by 'the membership'. To attempt to sell Troy as a closet GM level chess player is akin to suddenly presenting Worf as an accomplished concert pianist. Just can't get there from here. – Morgan Apr 9 '14 at 7:01
  • This newfound ability seems to be covered under the TV Tropes rules of engagement. Being a combination of; 'Required Secondary Powers' tvtropes.org/pmwiki/pmwiki.php/Main/RequiredSecondaryPowers 'Suddenly Always Knew That' power existed tvtropes.org/pmwiki/pmwiki.php/Main/SuddenlyAlwaysKnewThat and 'New Powers as the Plot Demands' tvtropes.org/pmwiki/pmwiki.php/Main/NewPowersAsThePlotDemands How did they suddenly acquire a new skill/ability? The writers simply add a new one if they feel that a new power/skill would open up a new storyline or puff up the character. – Morgan Apr 12 '14 at 6:26
  • 5
    Notably, this isn't chess but 3D chess. As we see in TOS, Kirk beats Spock by using an aggressive strategy. The implication is that emotion benefits playing ability. – Valorum May 18 '14 at 0:08
  • 3
    I agree. This is equivalent to all those poker scenes in Hollywood movies, where the hero is spectacularly beating his opponents with very unlikely settings. See Casino Royale for example. In the final hand, where (conveniently) all 4 remaining players are all-in, Bond has a straight flush while his opponents have 4-of-a-kind and 2 full-houses. This may be spectacular for the non-poker viewers, but utterly stupid for a professional poker player. For the latter, a spectacular win would be either a well executed bluff with a polarized range or a marginal call against a polarized range's bluff. – Bogdan Alexandru Jan 27 '15 at 16:39
  • @BogdanAlexandru Bond won by using 00-Level slight-of-hand. J/K – Gorchestopher H Aug 3 '16 at 16:11
73

They are playing three-dimensional chess, which may involve many more permutations for a small number of moves than ordinary 2D chess--even the best computer programs still can't typically beat professionals at the game of Go for example, because the number of permutations is much larger than in chess.

We also don't know if Data was programmed to do brute-force searches through vast number of combinations of moves like modern successful game-playing programs do, or if as an A.I. designed to mimic aspects of human thinking he was programmed to use more humanlike search strategies (which presumably requires intuitive thinking to decide which strategies seem more promising to focus on) even if it made him less successful (perhaps he was capable of both types of strategies, but chose to use the latter when just playing games as opposed to life-and-death situations).

Also, what makes you think Troi is a novice? The fact that she hasn't been shown playing 3D chess in previous episodes doesn't imply she never does, the characters could have plenty of hobbies that aren't shown. And to say she's shown "no particular skill at anything logical in her entire existence" also seems unwarranted, her job as Counselor may have favored intuition over logic but there was nothing on the show to suggest she was actually bad at logical thinking.

For some evidence that she had plenty of skill at logic and math, note that in "Thine Own Self" she did pass the Bridge Officer's Test which involved many stages, and when she at first failed to pass the final stage Riker said "Don't feel bad. You passed everything else. Diplomatic law, first contact procedures, Bridge operations. The Engineering qualification's one of the toughest parts of the test." So, it seems reasonable to assume some of these earlier stages involved plenty of technical knowledge and logical reasoning, especially the engineering qualification.

Also, getting into Starfleet itself required passing the Starfleet Academy Entrance Exam which involved things like a "hyperspace physics test" and solving a complex geometric problem (as depicted in "Coming of Age").

Improve this answer
15
  • 2
    Granted my comment about Troi's "no particular skill at anything logical" may be too specious. Be that as it may, her forte is intuition, empathy and reading other's emotions. She has shown no special degree of skill required to, as Data notes "devised a completely unanticipated response to a classic attack." Now, a "classic" attack in chess usually earns that moniker because it survives decades if not hundreds of years of not being exposed to a forced mate that even a computer can't defend. Is this new previously unknown ability a Trope Namer or just a writing error? – Morgan Mar 30 '14 at 17:17
  • 2
    @Hypnosifl It'd mean that in every single game where that move was made, there were two game-changing errors made (first by making a move where such a defeating counter-move exists; second by opponent not making that countermove) in a game where usually every possible move is considered; and that nobody ever noticed it in relaxed post-game analysis (that would naturally consider all possible deviations at least for the single move); and that it happened despite the situation happening often enough to have a commonly known name. – Peteris Mar 30 '14 at 18:17
  • 2
    During the last years AIs have made big progress on Go, so I think it's not realistic to think that by the time Star Trek is set, Go hasn't been broken yet. – o0'. Mar 31 '14 at 16:06
  • 8
    As of 2016, computers can beat humans at Go. It would seem logical that they can do the same for 3D chess by the time Data is created. – hkk Mar 26 '16 at 23:10
  • 4
    @troglodite To be fair though, we haven't even invented 3d Chess yet. Go has been around thousands of years for strategies to be developed that algorithms were created from. 3d Chess may be just a complex and we haven't even begun to play it yet, much less master it. – Prof. Bear Apr 7 '16 at 16:00
66

Because Troi is not playing the way Data expects

From the transcript:

(a 3D chess game is in progress)

DATA: The characteristic response to the Kriskov Gambit is to counter with the el-Mitra Exchange, particularly since I have already taken both your rooks. By missing that opportunity, you have left your king vulnerable.

TROI: We'll see.

DATA: As you wish, Counsellor. Check.

(Troi makes her move)

DATA: Intriguing. You have devised a completely unanticipated response to a classic attack. You will checkmate my king in seven moves.

TROI: Data, chess isn't just a game of ploys and gambits. It's a game of intuition.

DATA: You are a challenging opponent, Counsellor.

TROI: Thanks, but don't think a compliment's going to get you out of our bet. You owe me one Samarian Sunset made in the traditional style, as only you can make it, Data.

DATA: I will honour our agreement.

Couple things to note:

  1. I would not call Troi a novice. She is confident she is about to beat Data and has enough experience to form an opinion on chess. She is not surprised when Data calls out the number of moves she is going to beat him, they both know exactly what will happen. This wasn't blind luck, she lured Data into some kind of trap.
  2. Data's chess ability is obviously constructed from known classical moves. Troi is probably unaware of these moves completely, but is a competent enough player without them. This leaves Data somewhat blind.

However, if I were Troi, I would not expect Samarian Sunsets in the future. Data has probably memorized all of the moves Troi is making and would plausibly be able to predict them.

Improve this answer
10
  • 4
    Agreed. It's because he's very predictable. She knows what the traditional response is to her gambit, then subverts this by doing something different. – Valorum Mar 30 '14 at 17:10
  • 42
    It's simply a contrived movie plot which is incompatible with reality. I mean, if X is called a "characteristic response" with a given name; then it has been analyzed and played at least a few dozen times [conservatively] in the history of the game. There is a reasonable number of possible moves at any given position (which is true here). It naturally implies that all those moves have been considered; and if it has a possible countermove that leads to 'checkmate my king in seven moves', then it would be universally known as a classic bad move. Chess has no possible "unanticipated responses". – Peteris Mar 30 '14 at 18:11
  • 8
    I think there is merit to the reasoning that Data is himself learning Chess and applying his own interpretations. Given that the Kriskov Gambit and el-Mitra exchange are made up moves for a made up game, who is to say that there aren't a number of board configurations from which they could be played? Perhaps Data has already used an unconventional move or combination of gambits to reach an as-yet under-analysied board configuration and is therefore able to be lured into the trap. This proves Troi's competence and improvisation ability in 3-D chess – Dacio Mar 30 '14 at 20:46
  • 6
    @fredsbend, As Data has done with other subjects he has endeavored (acting, poker, Sherlock Holmes, etc.), he always accesses and downloads every scrap of information available pertaining to that subject. It's unlikely he simply read 'A' book about the subject of chess. He would more likely have read them all, as well as all analyses of every game ever played and all chess engine functions in existence. That's how he rolls. – Morgan Mar 31 '14 at 23:20
  • 5
    @Morgan By the same caveat, when he wants to act those things out he tends to take everything literally, like solving an already well-known Sherlock Holmes case by knowing how it will turn out, or simply combining all the elements of previous cases into one terrible mish-mash, before Geordi finally talks him into letting the holodeck create an original case for them to solve. He's good at learning, but has trouble with applications outside the standard. – Zibbobz Apr 1 '14 at 13:16
40

Data often strives to emulate humans - this likely goes as far as to try and play chess like a human rather than like a computer.

The episode "Peak Performance" actually goes into this; Data is soundly beaten at the game Strategema by an expert. Data is given advice to stop trying to play like others do and instead use his own strengths - in the rematch, Data plays much more like a computer, relying on his computational power, endurance, and level head to play extremely sound tactical defense, playing to avoid the possibility of defeat rather than seeking to win.

This ultimately leads to victory: the expert becomes frustrated by his inability to break Data's defenses and to lure Data into giving up a strategic weakness, and eventually rage quits.

Improve this answer
6
  • 12
    +1 for rage quitting in the 24th century. That episode ending is one of my favorites. The only problem is that Peak Performance is episode 2x12 and Conundrum is episode 5x14. So in that respect, it's still a plot hole (or another improbable failing on Data's part). – Dacio Mar 30 '14 at 20:42
  • 4
    Agreed. It's my feeling that Data had probably agreed to work via heuristics rather than brute-force search as a means of exploring his humanity (and because Troi wanted to sucker him into the bet...) – FuzzyBoots Mar 31 '14 at 0:52
  • 7
    @Dacio Also, the problem with episode 2x12 is that Pulaski is in it. – frеdsbend Mar 31 '14 at 19:58
  • 1
    @Morgan: I'm actually surprised that Data hadn't understood the ramifications of his strategy: "The good fighters of old first put themselves beyond the possibility of defeat, and then waited for an opportunity of defeating the enemy" (-Sun Tzu). The parallel with actual chess history is interesting too: one of the first major strategic advances in modern chess strategy was a focus on avoiding positional weaknesses. (such as ones that might arise from overextending themselves to pursue a trap disguised as an obvious avenue of 'advancement') – user12616 Apr 1 '14 at 8:24
  • 7
    I think that those concentrating on Data as a computer with some AI on it are missing the point. Data is first and foremost an AI, from his programming to his hardware. He's not a general purpose computer with a brute-force chess-playing program alongside some human-ish AI. His conscious mind may well work like a human who has perfect recall and is very fast-thinking, but like a human none-the-less. For better or worse. So for purposes of a chess game, he only has about a 7-move look ahead. (And he may have also underestimated how far he would have to look ahead to beat a touchy-feely.) – Wayne Apr 1 '14 at 22:06
23

I would say that Troi is a strong player using her logical thinking and intuition but who was also extremely lucky.

She surprised Data with her 8-move-long winning sequence but then it instantly detected checkmate in 7 moves.

Data probably was taking into account known gambits and documented game plays but couldn't use them to determine looming defeat as Troi created a new / undocumented position. That strongly suggests that a 7 moves deep search was used, which would simulate for possible follow-ups and leave no further surprises.

That means that Troi pulled out the shortest winning sequence outside Data's search scope. If she had an equally unexpected, brilliant move but leading to the victory even a tiny bit faster, Data would be aware of it thanks to 7 moves deep search.

That was light technical analysis but as Wikipedia states:

No complete solution for chess in either of the two senses is known, nor is it expected that chess will be solved in the near future.

The message here was that some problems aren't suitable for computers even if they have outstanding computing power. Intuition makes us superior to machines as it allows to simplify complex analytic problems.

Note that in 2D chess usually

  • every other pawn can make 1 move
  • knights, bishops, rooks and the king can make ~4 moves
  • queen can make ~8 moves

So in the middle of the game there are usually around 20 possible moves. That means that if X move deep search would take 10 seconds, X+1 move deep search would take ~66 hours (60s * 20 * 20 as there are two sides in the game). In 3D chess there are probably many more pieces and many more moves, so X+1 move deep search could take days or even weeks.

Improve this answer
2
  • 3
    I'm a chess aficionado myself, and I find this to be an incredibly good first post. I don't necessarily think the answer is correct, mind you, but it's a shockingly good first post. Well done. – James Sheridan Apr 1 '14 at 9:36
  • 1
    Thank you @JamesSheridan :) I'm an amateur chess player with computer science background so I thought my point of view might be worth presenting. It's based on a few assumptions but it nicely aligns with Staniław Lem's view on AI and mostly stays true to theoretical findings and techniques in use even though many algorithms for playing chess are actually randomized. – Legat Apr 1 '14 at 10:12
11

If I recall correctly, there is an episode in the original Star Trek series where Kirk is court-martialed for having authorized some fatal action while the required pre-conditions (namely, red alert status) were not fulfilled. Evidence is provided by the Enterprise's main computer. Spock then proceeds to demonstrate that the computer has been altered, by beating it repeatedly at 3D chess. Spock claims that if the computer was sound, beating it would be impossible (even from him).

Maybe they were playing a different variant.

Improve this answer
5
  • The main computer ought to have several orders of magnitude more capability than a mobile power-limited unit, no? – Ben Voigt Mar 31 '14 at 16:52
  • 2
    Yeah, but the Kirk/Spock era was a few decades before. Has technology stalled in the Star Trek world ? Here on "Real Earth" we can wear as wristwatch systems which are way more powerful than building-sized computers from a mere 35 years ago. – Thomas Pornin Mar 31 '14 at 16:58
  • 1
    @ThomasPornin: Yes, but we're at the dawn of the electronic era and our advancement has been exponential. While technology in ST is still improving, it's very unlikely that computing capability is improving at anything like the rate it does now, because they'll already be close to fundamental limits. (Heck, we're already hitting fundamental physical limits.) – Tynam Apr 1 '14 at 23:52
  • Of course, it's possible Spock knew this was wrong, and that he could beat the Computer with no problem, but produced this 'evidence' because he had good confidence that James T. Kirk wouldn't break the rules, and so had to open up an avenue by which the only logical culprit - the computer - could be investigated. – Zibbobz Apr 4 '14 at 17:17
  • It follows popular thinking. Initially it was thought that computers would crush humans, so Spock's program is unbeatable (if not tampered with). But computers were slow and easily beaten, which killed the early optimism, which is why Troi can beat Data. And now we're in the Stockfish era where we know that computers are better. – James Hollis Jul 5 '18 at 20:14
6

Computers usually just make better guesses than people in chess. The game is not solved, and it seems like it is not solved in Star Trek either. Computers use much more computational power to play, but unless they actually solve the game, there is no guarantee for a move being perfect. P.S. I'm into chess engine programming, I can explain in much more details, but you can also just Google for computer chess. You can get lucky while playing today's strongest chess program as well, but probably you won't often. Maybe once in a hundred years, this once in a hundred years of course happens more often on TV shows.

Improve this answer
3

Well, I wouldn't compare Data to supercomputers in the context of both having similar anatomy, but the way they are programmed to work. Supercomputers known to beat the world's best chess player calculates moves based purely on the best logical permutation of the move. It analyzes the strategies used and finds the best course of action.

Data has not been programmed to do what supercomputers are known to do. Data analyzes, understands, reads human emotions, and stores things he observes. He may not have all the answers, and most of the conclusions he has about human nature have been wrong; this is due to the show's tendency to exemplify how human emotions can not be understood no matter how advances the technology becomes.

So when Data plays chess, he doesn't compute permutations of the chess game, what he tries to do instead is read his opponent, which he is unable to do so.

To me, the question is like comparing electronics.. like a toaster to a camera.. and not wondering why a camera (though more complex) doesn't toast bread.

Improve this answer
3
  • All Data would need to beat most anyone below a Class A or Expert level player is to simply download a game database. Data is made for such a function. Most people can't go more than 5 or so moves into an opening sequence by memory before getting lost in the briar patch. You simply can't play chess by 'reading emotion' any more than you can perform math equations by reading emotion. – Morgan Apr 4 '14 at 17:50
  • 1
    Yeah, but Data is always curious about human emotion and since he plays against a human, I'm suggesting that he believes that all decisions are based on reading humans. Like Data has no concepts of bluffs. For humans, sometimes we can't see everything, like a bishop targeting a queen from the other side of the board, those factor in as the "unseen"; if that is missed, Data might be curious as to why we didn't see it.. and might consider it as a bluff. Data might be extremely knowledgeable, but he is humble in knowing that there are things he doesn't know. – sksallaj Apr 7 '14 at 17:06
  • To further expand on the comment by @Morgan, even if you download the game database to calculate the permutations, that can still be interfered with Data's other functions of reading emotions. Even if he has the answers to beat his opponent, his interactions with humans may lead him to doubt his intentions. – sksallaj Oct 9 '15 at 18:55
3

(Troi makes her move)

DATA: Intriguing. You have devised a completely unanticipated response to a classic attack. You will checkmate my king in seven moves.

Data was able to detect that Troi will checkmate his king in 7 moves. This means that from this position, no matter what move Data makes he will lose in 7 moves. Data could have only said this after analyzing all the paths in the game tree from this position. We do not know how much time Data took to come to this conclusion. But it was a finite amount of time and would have been quick. (Or the transcript would have said, "after a long pause" Data replied).

The game tree at this position could have been very big, but Data is also powerful. Let's not worry about the numbers here. The fact that Data was able to find the Mate in 7 scenario, it should have found it before the previous move. When Troi had to move she would have got about roughly 200 choices to make. Given that the game was going on classical line up to this point, Data would have analysed these 200 choices upfront and if one of them is getting it into a Mate in 7 scenario, then this will never be a classical variation.

Note: I think this argument will not make sense if some one can prove this statement

In any given position a Mate in 8 cannot be computed (using a polynomial time algorithm) given the fact that one of the next move results in a Mate in 7 which can be computed using a polynomial time algorithm.

Improve this answer
2

I don't know much about Data, but I do know about AI. Data was likely created for more general tasks and intelligence, and not specific ones like chess. Most chess AI's are created specifically for chess, and therefore excel at it. Comparing Data to a chess AI may be like comparing a Xbox to a computer. A Xbox can display some highly detailed environments, but that is because it isn't worrying about updating the time, downloading your video, and running your other programs for you. Just like Xboxes can do one thing better then a computer, a chess AI can do chess better than Data.

Improve this answer
4
  • 4
    This is what I was going to say, too. Data is an artificial intelligence. That is a program running on a computer. Data isn't a supercomputer with a program for solving every problem in the universe built right in. We know that Data has to "create" programs for himself. He has to do that for romance, so why not for chess also? In which case, he is obviously still developing it at the point that he plays Troi, and he discovers that he needs to cover more than just common gambits... – GreenAsJade Apr 2 '14 at 6:43
  • I'd point out that there is little to nothing an Xbox can do that a computer can't do better, and for barely higher cost – The Fallen Apr 2 '14 at 22:42
  • Actually, an Xbox is just a fancy Windows computer. The earlier Dreamcast, running Windows CE with DirectX, had better effects than the PS2 thanks to mature development tools. Sony released a PS2 Linux distro for general purpose tasks. – Cees Timmerman Apr 4 '14 at 10:45
  • True, true, but Xboxes are not running as many processes at once and has more graphics processors. – user2035846 Apr 28 '14 at 4:35
1

I don't think she actually beat him at all. If you look at the move Data made, he appears to have put her in check with a rook (although it might be his queen). Then after her move, her King is still in check. Therefore, her move was illegal.

Improve this answer
1
  • As we don’t know how the geometry of 3D chess works, it is possible that she moved into the line of check (however “line” may be defined in 3D). – chirlu Oct 7 '17 at 12:44
1

It's not all about computation. Humans have superior understanding. Computers understand nothing but they have good algorithms. In chess, computation plays a big enough role for computers to win most of the time. Not so in other games, like possibly mahjong and poker. In poker, computation is definitely less important.

Improve this answer
2
  • 2
    What do you mean by "understanding?" That sounds like a handwave. – user1786 Apr 1 '14 at 23:09
  • @JonofAllTrades Probably inferring rules better thanks to experience. Knowing the mood of your opponent is not something a computer tends to be designed to do, and the ship scanning that in case it has to play poker later that day is probably taken as an invasion of privacy. – Cees Timmerman Apr 4 '14 at 10:53
0

Data can lose because he deliberately held back from using a brute-force solution, similar to how he could almost certainly mark cards in their weekly poker games, but doesn't. Data has no need to demonstrate perfect play on a solved game, except maybe at a math symposium; he doesn't seem to take any particular pleasure from it, and he won't learn anything from it. (This is assuming chess is a solved game in the 24th century.)

On the other hand, deliberately restricting himself to treating chess as an unsolved game and attempting to play it intuitively, like a human does, teaches him several things. It allows him to sharpen his skills at reading an opponent, it helps him to understand how humans reach the decisions they make, and it lets him apply the lessons he learns to other unsolved games like interstellar diplomacy or starship combat.

Improve this answer
4
  • Do you play chess? – Morgan Oct 29 '19 at 23:06
  • @Morgan About as well as Troi does, but it's not about chess. It's about Data learning to be more like a human being. He doesn't need to learn how to win at chess, he already knows that. – Cadence Oct 29 '19 at 23:29
  • I actually am a rated chess player... Are you contending that Data threw the game with Troy in order to 'be more human'? How does that work? – Morgan Oct 29 '19 at 23:39
  • @Morgan Not so much threw the game as handicapped himself, by treating it not as a test of how well he can play chess, but of how well he can imitate a human playing chess. – Cadence Oct 29 '19 at 23:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.