'A Most Unconvincing Replicant: Allegory and Intelligence in Blade Runner’s Chess Game', by Jacob Edwards

Blade Runner (dir. Ridley Scott, 1982) is set in a world rich with background detail. In terms of creativity—such as the complex ‘cityspeak’ fashioned at great length by Edward James Olmos (who plays the character Gaff), although it features only briefly—the film demonstrates considerable attention to detail. Yet, in one of its more mundane depictions—a game of chess—Blade Runner is found wanting: the game between Sebastian and Tyrell has been rendered with a level of intent and accuracy sufficient only to draw attention to its flaws.

The chess game performs a simple function in terms of plot—giving Roy a means by which to approach Tyrell—and allows for an easily perceptible ranking of Sebastian’s, Tyrell’s and Roy’s respective intellects. This may serve adequately for most viewers. Megan de Kantzow, author of the New South Wales high school study guide for Blade Runner, for instance, accepts (p. 86) that Roy gained access to his creator because Tyrell was “intrigued” by Sebastian’s move. However, a more detailed examination reveals that the specifics of the chess game do not align well with its apparent function within the film.

Blade Runner has seen several different releases, but none of these versions differ in their portrayal of the chess game. Let’s first recap, superficially, the relevant sections of the film. Sebastian and Tyrell are playing chess by correspondence. The game features in two scenes:

Scene #1 (from the Director’s Cut, 73:24 to 75:41)

Roy catches sight of the chess board in Sebastian’s apartment. After studying the position for just under ten seconds, he makes a move. Instantly, Sebastian corrects him.

Sebastian:      No. Knight takes queen. See? No good.

Roy moves to the other side of the board, plays Sebastian’s supposed refutation, and then ponders the new position. The focus of the plot shifts for a few minutes before returning to the game.

Roy:                 Is he good?

Sebastian:      Who?

Roy:                 Your opponent.

Sebastian:      Oh, Doctor Tyrell? I’ve only beaten him once in chess. He’s a genius.


Scene #2 (from the Director’s Cut, 77:48 to 79:02)

Roy and Sebastian are taking the elevator up to Tyrell’s penthouse. The lift stops and Tyrell questions Sebastian over the intercom.

Tyrell:             What can I do for you, Sebastian?

Sebastian:      Queen to bishop six, check.

Tyrell:            [Moving from his bed to the chess board] Nonsense! Just a moment. Hmm. Queen to bishop six. Ridiculous. Queen to bishop six. Hmm … Knight takes queen. What’s on your mind, Sebastian? What are you thinking about?

Roy:                 [Whispering] Bishop to king seven, check-mate.

Sebastian:      Bishop to king seven—checkmate, I think.

Tyrell:             Got a brain-storm, huh, Sebastian? Milk and cookies kept you awake, huh. Let’s discuss this.

The film industry may be notorious for its mis-depiction of chess—few serious chess players would be surprised that the position on Sebastian’s board in no way matches that of Tyrell’s—but in Blade Runner’s case there seem to have been some pains taken to portray a real, historical game. From the specificity of the moves played, and what can be made out from Tyrell’s board (the clearer of the two), it can be seen that Sebastian and Tyrell are re-enacting Anderssen versus Kieseritzky (London, 1851), an encounter that has been dubbed by history as the ‘immortal game’.

To reach the position that Roy first contemplates, Anderssen and Kieseritzky played the following moves: 1.e4 e5 2.f4 exf4 3.Bc4 Qh4+ 4.Kf1 b5 5.Bxb5 Nf6 6.Nf3 Qh6 7.d3 Nh5 8.Nh4 Qg5 9.Nf5 c6 10.g4 Nf6 11.Rg1 cxb5 12.h4 Qg6 13.h5 Qg5 14.Qf3 Ng8 15.Bxf4 Qf6 16.Nc3 Bc5 17.Nd5 Qxb2 18.Bd6 Bxg1 19.e5 Qxa1+ 20.Ke2 Na6 21.Nxg7+ Kd8


Starting from this point, Roy gives—in the obsolete descriptive notation that Anderssen and Kieseritzky would have used, rather than the algebraic notation that is prevalent in modern times—22.Qf6+ Nxf6 23.Be7 (checkmate)

William M. Kolb, in his ‘Blade Runner Film Notes’ (pp. 165-166), observes that Sebastian is scripted in the film as a chess grandmaster, and that this both highlights Tyrell’s genius and elevates Roy to an altogether new level of intelligence. However, all three players act in a manner that is thoroughly inconsistent with this portrayal.

Firstly, Sebastian. His response to Roy’s first move—“No. Knight takes queen. See? No good.”—is that of a beginner to intermediate level player, dismissing the move simply because the queen can be captured. In reference to the whisperings that Ridley Scott dubbed onto Sebastian’s automatons during post-production, Paul M. Sammon, author of Future Noir, suggests (p. 169) that Sebastian’s “No. Knight takes queen” comment is ‘reinforced’ by a ghostly murmur of “…will mate you.” Logically, however, “…will mate you” can refer only to the positive outcome of Roy’s move, not to Sebastian’s specious refutation. In this case, it seems that even Sebastian’s (admittedly perspicacious) toys can see the merit of Roy’s play. Given the precarious position of Tyrell’s king, and that Sebastian already has sacrificed (or blundered—inconceivable for a grandmaster) two rooks and a bishop, he would actively be seeking moves just such as this one. Strong players rarely are as shallow in their thinking as Sebastian appears, and in the historical game Kieseritzky resigned after 20.Ke2 (or possibly 20…Na6 with Anderssen having announced the winning continuation). Perhaps the original intention had been to start Roy’s calculations one move earlier—the queen sacrifice is more difficult to see with the black king still on e8, and with a white knight on f5 denying the queen access to f6—but although this is more credible, the innocuous move ‘knight takes pawn, check’ could hardly give rise to the dismissive response that the script requires of Sebastian.

Secondly, Roy. Although his initial consideration of an unfamiliar position is natural, his subsequent behaviour does not match that of a strong player. The two moves—queen to bishop six, check, and bishop to king seven, checkmate—are bound together. The first, which serves to decoy Tyrell’s knight from g8, is played solely to set up the second. Sebastian’s supposed refutation merely confirms that Tyrell would have no choice in his next move. Roy, having seen the two move combination, would have no need to doubt his own analysis or reconsider the position. Charitably, it could be argued that he moves to the other side of the board so as to put himself in Tyrell’s mindset—but this is a purely symbolic interpretation of what, in terms of the chess game itself, remains inexplicable behaviour. Roy’s question to Sebastian—“Is he good, your opponent?”—also makes little sense if Roy is assumed to be intelligent enough (as someone beyond-higher-than-grandmaster level obviously must be) to evaluate the position independently. Even if Roy understands nothing but the rules, and is progressing from beginner to super grandmaster in a matter of seconds, it would be more appropriate to ask, “Is he considered good, your opponent?”

Thirdly, Tyrell. For a supposedly higher-than-grandmaster chess player, Tyrell is unnaturally flustered by Sebastian’s move. He cannot seem to visualise it away from the board, and even when contemplating it at the chess set, he does not immediately appreciate its significance. Admittedly, his mind may be racing ahead in cogitating the non-chess subtext of the game, but to merely play ‘knight takes queen’ without acknowledging the inevitable checkmate, is more suggestive of an amateur player than a preoccupied genius. In propounding that the chess game elevates Roy above Sebastian and Tyrell, William Kolb notes (p. 175) that it is not unheard of for experts to overlook a two move checkmate, and even cites Rubinstein versus Nimzowitsch (San Sebastian, 1912) as an example. However, although the chess dynamic between Sebastian and Tyrell may be as Kolb describes, and likewise the significance of Roy’s contribution, the ‘brilliancy’ of the two moves nevertheless has been misappropriated. The contestants themselves are depicted as (at most) something akin to social club players.

Further to the character-players not acting the relative parts of grandmaster, greater-than-grandmaster and somewhere-beyond-super-grandmaster, the game itself is not appropriate for conveying this dynamic. Adolf Anderssen and Lionel Kieseritzky were very strong players of their age, but chess in those days retained a significant romanticism that both encouraged sacrificial play and frowned upon the non-acceptance of gambits. The ‘immortal game’ is so called because it perfectly captures the spirit of that age. Anderssen sacrifices a bishop, then both rooks and finally his queen, before checkmating Kieseritzky using his remaining three minor pieces. Kieseritzky, though having lost only three pawns, has not been able to develop his pieces properly, nor move his king to safety. He is powerless to prevent checkmate. It is, without doubt, an unforgettable game, and one that most beautifully demonstrates the see-sawing balance that exists between strength and mobility. It was also, however, a friendly game contested during a break between rounds of a serious tournament. Anderssen’s play is whimsical as much as inspired, and Kieseritzky’s defence (if it may be called that) is little more than an indulgent acceptance of the material on offer. Writing the series My Great Predecessors in 1993, when he was world champion, grandmaster Garry Kasparov commented (vol. 1, p. 25), “Objectively the game is rather weak and superficial, but what a finish!”

The quality of play makes Anderssen versus Kieseritzky—and simply its re-enactment in Blade Runner—seem poorly chosen as an attempt to characterise Sebastian, Tyrell and Roy through their respective intelligences. The flow of the game does likewise. In the film the viewer supposedly is presented with a very strong player (Sebastian) matched against an even stronger player (Tyrell) whom he has beaten only once. Sebastian is headed for another defeat until the intervention of Roy, whose intelligence burns “twice as bright”. This is a difficult scenario to achieve, particularly in a correspondence game, because it requires a seemingly inferior position to be turned, via some innate piece of brilliance, into a winning position. An oversight made in time trouble—such as happened in the game to which Kolb refers between Rubinstein and Nimzowitsch—would not suffice; nor even would a more sublime queen sacrifice such as in Levitsky versus Marshall (Breslau, 1912):


Marshall won by playing 23…Qg3! This move looks utterly outrageous at first—the queen can be captured by either of two pawns, or indeed by white’s queen—yet it leads to a forced checkmate. (The threat is 24…Qxh2 which is unstoppable in light of the forcing variations: 24.hxg3 Ne2#; or 24.fxg3 Ne2+ 25.Kh1 Rxf1#; or 24.Qxg3 Ne2+ 25.Kh1 Nxg3+ 26.fxg3 Rxf1#) Black’s winning move may be spectacular, but it arises from an already superior position, and relies for its checkmating combinations upon black’s extra knight. Well might the spectators have thrown gold coins onto the board, yet still this remained a powerful end to a game in which Marshall dominated, not a telling reversal of fortunes.

In the ‘immortal game’, Anderssen (Sebastian) very deliberately plays his way into the winning position, and Roy’s concluding queen sacrifice is nothing more than the logical—and only—extension of the preceding moves. Given that the flow of play in no way approximates the required contrast of intelligences, one cannot help but conclude that the ‘immortal game’ was chosen as much for its name as for its relevance to the human/replicant dynamic. This is perhaps further evidenced by the extemporised nature of the chess game’s inclusion. According to Sammon (pp. 55-56, 171), Hampton Fancher, who wrote the early Blade Runner scripts, originally had Sebastian spiriting Roy into the Tyrell “preserve” under the guise of delivering an artificial griffin (or miniature unicorn). Ridley Scott and producer Michael Deeley insisted on replacing this sequence with a less expensive plot device. In one of several instances of the artificial animals/electric sheep thread being pulled from Blade Runner, Ridley Scott instead conceived of the chess game, which then was scripted by David Peoples.

Given that Blade Runner’s inherent question—“What is human?”—would lose much of its meaning if not for the replicants’ almost palpable mortality, it seems intuitively wrong that there would be no significance in the act of two mortals—one human, one replicant—contesting the ‘immortal game’ of chess. Yet Ridley Scott, when asked whether Blade Runner’s chess game was structured in homage to Anderssen versus Kieseritzky, dismissed the idea, saying (Sammon p. 384):

“The answer is no. What’s that line you see at the end of film credits? ‘Any resemblance between this photoplay and actual events is purely coincidental?’ [laughs] I’m afraid that’s the case with Blade Runner’s chess game—it’s purely coincidental.”

It’s a glib reply, and unconvincing. Granted, not enough of Tyrell’s board is seen to establish conclusively that it shows the exact position of the ‘immortal game’, but the moves played—two for white, one for black—are highly distinct. Qf6+ is unusual in any regular game of chess, if for no other reason than because black’s king starts on e8 and usually relocates (via castling) to either g8 or c8, none of which squares are attacked from f6. If the king does find its way to d8 then almost invariably it is because the white and black queens have been swapped on that square, an act that again precludes the possibility of Qf6. The odds against white’s queen moving there, with check but without capture, are lengthened still further by the fact that the square is guarded. Indeed, the knight on g8 usually is developed quite early in the game, and for it to take any piece—let alone a queen—on its first move is virtually unheard of. Once the pièce de résistance of Be7 is taken into account—a minor piece checkmating in the middle of the seventh rank, attacking a square upon which the king hardly ever resides—the probability against the Blade Runner sequence being unintentional becomes astronomically high.

Further still, Ridley Scott was not likely to have let his crew sit twiddling their thumbs, playing an infinite number of chess games until there manifested a position of Shakespearean eloquence. Scott took complete control over every aspect of Blade Runner, noting in one interview (Sammon p. 387) that, “to put that kind of thing on screen requires enormous attention to detail. And it can finally only be accomplished through one pair of eyes.” Probabilistically, there can be no coincidence. Having signed on as director, Scott personally oversaw all aspects of script development, and the moves scripted for Sebastian, Roy and Tyrell leave no doubt: the chess game in Blade Runner is Anderssen versus Kieseritzky.

So, somebody set up the chess boards. Somebody—presumably at this stage David Peoples—wrote the exact words of dialogue. Perhaps Ridley Scott genuinely suspected nothing of the connection between the ‘immortal game’ and that of Sebastian/Roy versus Tyrell, but in that case it might well be argued that he should have known, and that the chess game should have been either better chosen or specially constructed, if not to characterise Sebastian, Roy and Tyrell through means of their moves, then at least not to undermine their characterisation (as does the current sequence). Would it have been possible to stage a game more apposite to the participants’ supposed intelligences? The answer is yes, and the key to such a scenario lies—fittingly enough—in an appreciation of the difference between artificial and human intelligence. Consider the following position from game two of the 2006 match between computer programme Deep Fritz (white) and reigning world champion Vladimir Kramnik (black):


The position is even—black with the advantage, white set to draw by perpetual check—but Kramnik, with half an hour left on his clock, having planned his continuation six moves previously and checked it on every subsequent move, stunned the world by playing …Qe3?? and thus allowing Fritz to checkmate him in one move with Qh7.

In contrast to Tyrell’s disbelieving bluster, Kramnik reacted calmly to the ‘surprise’ nature of his opponent’s move, merely raising one hand to his forehead before sitting down, signing the scoresheet and leaving to attend a press conference. But what was the cause of Kramnik’s monumental error—a move that grandmaster and blogger Susan Polgar called “Blunder of the century? Biggest blunder ever?” Kramnik himself could offer no excuse, but, given that one grandmaster commentating the game also overlooked the checkmate, Alexander Roshal (editor of Russian chess magazine 64) proffered the following explanation: that the positioning of white’s knight on f8 is so unusual as a means by which to support a checkmate on h7 (almost always it would be on f6 or g5) that Kramnik’s brain, his organic pattern recognition software, simply did not register the threat.

Herein lies the intrinsic difference between man and machine in their respective approaches to chess. Grandmasters are highly imaginative and can calculate to an extraordinary depth, but their calculations are filtered and directed according to positional assessments, which are based on experience. There is, in short, such a thing as ‘chess memory’. Strong players rely not only on strings of individual moves but also on the ‘chunking’ of analogous blocks of information; and whereas computers may be programmed to mimic this apperceptive intuition—that is, to evaluate a position either more or less favourably in instances where human experience has shown the benefits or drawbacks of a particular type of move, in a particular type of position, to be beyond strictly computational range—a computer retains its processing power at all times, and does not suffer from chess ‘blindness’. No chess computer, not even the simplest of mobile phone applications—a far cry from artificial intelligence—would ever make Kramnik’s blunder. Any half-decent programme would have played Roy’s combination within seconds.

What the Blade Runner chess game attempts, but fails, to depict, is an ability in Roy to find, through computation, a winning line that lies hidden to the grandmasterly experience, intuition and imagination of both Sebastian and Tyrell. This could be achieved, realistically, by taking some of the familiarity—the ‘chunked’ patterns—out of the game. In 1996 former world champion Bobby Fischer formally announced the rules for Fischer Random chess, which in essence, by shuffling the pieces on each player’s first rank, allows for 960 different starting positions. Fischer’s contention was that knowledge had begun to outweigh ability in chess games at the top level, and that the game had to be de-theorised if it were to become again a true test of intellect. He even demonstrated this, believes chess player and journalist Tim Krabbé (Open Chess Diary, entry no. 139), by playing anonymous blitz (three minute) games against strong opposition on the internet, making strange, almost random moves in the opening, and then fighting back to win from objectively worse positions because he was able to outplay his opponents amidst the resulting unfamiliarity. Such is the type of scenario that would have been necessary to convey Roy’s inherent superiority over Sebastian and Tyrell. Imagine the following position, arising from a game of Fischer Random chess:


(composed by grandmaster David Smerdon)

Scene #1 (with the chess board appropriately set)

Roy catches sight of the chessboard in Sebastian’s apartment. After studying the position for just under ten seconds, he makes a move: Qd3+. Instantly, Sebastian corrects him.

Sebastian:      No. King takes queen. See? No good.

Roy moves to the other side of the board, plays Sebastian’s supposed refutation, and then smiles. The focus of the plot shifts for a few minutes before returning to the game.

Roy:                 Is he considered good?

Sebastian:      Who?

Roy:                 Your opponent.

Sebastian:      Oh, Doctor Tyrell? I’ve only beaten him once in shuffle chess. He’s a genius.

Scene #2:

Roy and Sebastian are taking the elevator up to Tyrell’s penthouse. The lift stops and Tyrell questions Sebastian over the intercom.

Tyrell:             What can I do for you, Sebastian?

Sebastian:      Queen to queen three, check.

Tyrell:            [Moving from his bed to the chess board] Nonsense! Just a moment. Hmm. Queen to queen three. Ridiculous. Queen to queen three. Hmm … King takes queen. What’s on your mind, Sebastian? What are you thinking about?

Roy:                 [Whispering] Castles queenside, check-mate.

Sebastian:      Castles queenside—checkmate, I think.

Tyrell:             Got a brain-storm, huh, Sebastian? Milk and cookies kept you awake, huh. Let’s discuss this.

In the position that grandmaster Smerdon has devised, black (Tyrell) is two pawns up and, although his own king is precarious, is threatening both white’s queen and a revealed attack on white’s king.  White (Sebastian) appears at first to have a strong continuation in 1.Nxe4 but after 1…Qxe4+ 2.Qxe4 Bh7 3.Qxh7 g6 4.Bxe3+ Kd1 there is nothing that he can do to prevent 5…c2 (checkmate)

The dynamic perfectly fits the characters, yet still leaves room for Roy to find 1…Qd3+ 2.Kxd3 0-0-0 (checkmate). The winning combination, deceptively simple, is easy to overlook because the rules of Fischer Random chess, unlike regular chess, allow the king to start on—and therefore castle from—a square other than e1 (white) or e8 (black). In this case, white’s king would move right one square to c1, with the rook lifting over to d1 to deliver the checkmate. Such is the nature of this surprise move that it would even allow for Tyrell’s flustered response.

As GM Smerdon has demonstrated, then, it would have been possible to show a scenario wherein a high intellect (Sebastian) is elevated beyond a higher intellect (Tyrell) though the intervention of an intellect that is even higher still (Roy). Doubtlessly, the complexity of such a chess position and winning moves would have rendered the game incomprehensible to most viewers—that should, in fact, be the point—but its function within the film, and the impression it gives of Sebastian, Tyrell and Roy, would have remained the same. Not only would the use of a Fischer Random game (or ‘shuffle’ chess, as early variants were called) have carried a more futuristic feel, it would have allowed the scene to withstand the scrutiny of chess players and non-chess players alike.

Although Blade Runner was made before the advent of Fischer Random chess, and before Kramnik’s famous blunder against Deep Fritz, the concept of ‘shuffle’ chess was already well established by 1982, as was research into how grandmasters think. The truth of the matter is that the visionaries responsible for Blade Runner—and ultimately, the blame must fall here upon Ridley Scott—though looking forwards in almost all other respects, chose inexplicably to look backwards when scripting the chess scenes. The resultant depiction—intellectual heavyweights clumsily re-enacting simplistic moves from a century and a half earlier—not only comes across as implausible and amateurish, but also was eminently avoidable.


References and Further Reading

Megan de Kantzow, Brave New World and Blade Runner: The Director’s Cut (Pascal, 2001)

Garry Kasparov, My Great Predecessors (Gloucester/Everyman, 2003)

William M. Kolb, ‘Blade Runner Film Notes’, in Kerman, Judith B., ed., Retrofitting Blade Runner: Issues in Ridley Scott’s Blade Runner and Philip K. Dick’s Do Androids Dream of Electric Sheep, second edition (University of Wisconsin Press, 1997), pp. 154-177.

Tim Krabbé, ‘Open Chess Diary 121-140’, 9 June - 26 September 2001. http://www.xs4all.nl/~timkr/chess2/diary_7.htm

‘Man vs machine shocker: Kramnik allows mate in one’, ChessBase News report, 27 November 2006. http://www.chessbase.com/newsdetail.asp?newsid=3509

Susan Polgar, ‘Blunder of the century? Biggest blunder ever?’, Susan Polgar Chess Daily News and Information, 27 November 2006. http://susanpolgar.blogspot.com/2006/11/blunder-of-century-biggest-blunder.html

Alexander Roshal, quoted in, ‘How could Kramnik overlook the mate?’, ChessBase News report, 29 November 2006. http://www.chessbase.com/newsdetail.asp?newsid=3512

Philip E. Ross, ‘The Expert Mind’, Scientific American 295.2 (August 2006), pp. 64-71.

Paul M. Sammon, Future Noir: The Making of Blade Runner (Orion, 1996)

Eric van Reem, ‘The birth of Fischer Random Chess’, 31 May 2001. http://www.chessvariants.com/diffsetup.dir/fischerh.html


Jacob Edwards lives in Brisbane with his wife and son, and may be found online—conspicuously not blogging—at www.jacobedwards.id.au


This piece is from our most recent print issue, TLB20. You can grab a copy for yourself!