sábado, 12 de novembro de 2011

Players Chess News (1986) - A History of Chess Computers (parte 5)

Todos os créditos desta matéria pertencem a Players Chess News (1986) com atualização de Source: 11-1986, International Players Chess News: A History (Recap) of Chess Computers.
(Slightly edited by Hein Veldhuis.)


Early 1985
The latest development in chess computers for 1985 is Fidelity’s Chess Challenger 12 running at 5 Mhz. The older Chess Challenger 12 ran at 3.25 Mhz, but both machines use the same program, the Budapest. The increased speed of the Chess Chalenger 12 does not make it stronger than the Elite 4.0 which has the Glasgow program, a one year further advanced program. In other words, the increased program strength of the Elite 4.0 outweighs the greater clock speed of the Challenger 12. The new Chess Chalenger 12 increased $100 in price over its predecessor, from a list of $250 to the current list of $350. The rating strength increased approximately 50 points. Another new computer has arrived on the market, the Turbo Star from Scisys. Our computer experts are now testing this machine to evaluate its overall strength.

Late 1985
Until Fall, the Super Constellation was the strongest machine commercially available and should continue to offer the best price/performance ratio in the 2000 rating range into 1986. The latest models coming off the line have an even lower defect ratio than the initial models, which were already less than 5%. A larger version of the Super Constellation is in the works, called The Constellation Expert - expect a November delivery.
With an extra larger program than the Super Constellation (64K vs. 56K) The Constellation Expert will also feature substantial programming refinements. This should put in about 75 points higher than the Super Constellation, which has an official rating of 2018. Moreover, The Constellation Expert will be a full sized beautifully finished auto-sensory machine. But with a price-tag of $700, The Constellation Expert may have trouble meeting the sales of the popular Super Constellation, with its healthy price/performance ratio. Fidelity’s Elite A/S 4.0 remained a good buy as a beauty/brains model until the arrival of Novag’s Constellation Expert. Mephisto is back with two upscale programs, the “S” and Blitz, and five different choices for casings.
Their top-of-the-line model is the Munich S, a beautiful wood cabinet filted with costly electronica. The name “Munich” describes the 17” x 17” casing. The “S” stands for the expensive advanced program expected this Fall/Winter. Presenty, the Munich board is only available with the Blitz program. Although the “S” program is much larger than the Blitz with its 64K (and should break the 2000 harder), it is slower and hence better only at strategy. The Blitz is better at fast response, tactics, and especially problem solving. Better positional play should still put the Munich S as least 50 points ahead of Blitz. AVE Micro Systems recently released their Sargon 4.0 program. This program should not be confused with the Elite 4.0 or Prestige 4.0 program, although the casing looks very much like that of the Prestige (without the clock). The playing strength is now Class “A” (although claims over 2000 are made), a healthy class ahead of Sargon 2.5 or Sargon 3.0. The main reason for the jump into Class “A” is that the original programmers for the Sargon 2.5, the Spracklens, chipped back in to aid AVE Micro Systems. The rating is still a notch below Fidelity’s Elite 4.0 program due both to the sophistication of Fidelity’s program and its clock speed of 4 Mhz. However, with a program only slightly stronger than Fidelity’s new model costing 1/10th the price, few players will pay the premium for what is perhaps the most beautiful chess computer casing ever produced.





The dramatic development in late ‘85 is in the public arena - not in increascd playing strength (non tournament players, who constitute the biggest market, can count on loting every game against present top models and certainly don’t need anything stronger) but in improved price/strength ratio in the lower range, Fidelity’s new Excellence model would have been topof-the-line only a few years ago, doubtless carrying an expensive price-tag (the original plastic Elite listed for $1000). But the Excellence cost only about $100 with a perfectly decent playing strength of about 1800.
Chess computer manufacturers have a long history of overrating their machines. The only computer officially expert rated is the Super Constellation as 2018, yet Novag continue’s to claim a strength of over 2200. Such boasting is unnecessary since as its real strength it offers the best price/performance ratio for stronger players. Fidelity is guilty of the same exaggerations in the lower-end range with the Excellence, boasting a strenght about 200 points too high. Imagine trying to have an 1800-rated program running on inexpensive electronic components - the result could be even more quality problems than Fidelity’s expensive machines now experience. And it is not reasonable to expect the Excellence’s program which is only half the size of the Elite’s, to be as strong as that machine’s. It runs 25% slower than the Elite’s and contains only 60% of the opening moves. Also, the machine is not modular, and is incapable of accepting a large opening repertoire, or update.

The 3 Mhz clock speed is misleading since the program runs a cheaper processor yielding an even slower comparative speed. One advantage, however, of this procescor is the increased battery life: now the machine will play 100 to 200 hours on a set of four batteries.

Fidelity has plans to submit the Excellence to the USCF for an official rating, the second entrant to take this bold test after the Super Constellation. The hope is that the machine will come out with a rating over 1800, but that even if it fails significanrly short of that goal - say, low class “B”, It will still be by far the best buy for the dollar in the $100 price range, two other models are all below class “B”. The next models up, the Chess Challenger 12 and Constellation 3.6 both list for $250 and are at best only slightly stronger. Some impatient customers may still opt for the Novag Constellation 3.6 since it plays a much stronger speed chess game. Relatively speaking, the Fidelity Excellence is poor on fast response levels and needs 5 to 10 minutes per move for its strength to flourish. Other customers may opt for the handsome walnut-framed Chess Challenger 12 for cosmetic reasons, as the Excellence looks a hit like a home-made model kit.

Even the plastic Constellation series look very handsome, with a robust casing in which one can see the work of a highpowered team of designers. And the cost is not that much more - less than $100 difference after discounts (The Excellence lists for $225, plus $12.95 for the adaptor).
Stil, all in all, we feel the Excellence is the best buy for the dollar in the lower price range and will help anyone rated class “C” or below improve quickly. Ones one reaches class “B” or “A” the machine becomes more a sparring partner, useful for keeping the player on his toes.
In order to make full use of a chess computer as a tutor er self-improvement tool, its playing strength should be at least one solid class ahead of the player’s so that he can respect its advice and can plan on frequently losing and learning.



But even the non-tournament player, who can anticipate not winning on tournament time settings, will find the faster response levels easy enough to show him a win on occasion. So it makes a great gift even for the total novice (unlike the regular Constellation, which Novag should probably have built in training levels as they did with the Super since its number 1 level of 0-to-5 seconds response is not much worst than the slow response levels; on all levels the person en the street can count on losing).
Customers who cannot contend with the Excellence’s quality control have two options that were top-of-the-line models in their time, and which have been reduced substantially because they are used Great Game Machines with all three cartridges (over 1600 strength) $99.95 (including adaptor); or regular Constellations, rebuil
t $175, used $125, with used adaptors, $9.95.

Stronger players (class “B” and above) and lower-rated players ambitious to reach an Expert rating (the aspiring junior, for instance) will find their best buy in the Super Connie. Experts regularly use it to keep in form - even a few Masters have bought the machine, finding it sufficient to provide a real challenge, and enough of a sting to keep their interest. Masters can utilize the programmable opening book to steer themselves into middle-games that are part of their repertoire, but pose paricular difficulties.
To summarize, our recommendations for good choices in late 1985 are as follows: Fidelity Excellence - fine for the less ambitious player and as a gift for the casual player (e.g. most people.) Novag Super Constellation - appropriate for the serious tournament player looking to improve. And with its lower training levels, even an ambitious beginner has the whole spectrum ahead of him.
Mephisto Blitz - ideal for the problem solvist. Fidelity Elite A/S - for the person who appreciates beauty and brains. You could, patience notwithstanding (= onverminderd - ongeacht), wait for the Constellation Expert.
The matter of ratings It should be noted that Novag is the only company to have their computer officially rated by the USCF computer rating agency. Most companies won’t risk permitting their products to be officially rated because of the possible bad press if theirs does poorly, The Chess Federation allows any production or experimental models to play in regular rated human tournaments (eg. U.S. Opens). However, since they cannot monitor the legitimacy of “on-market” models, all computer entrants are considered experimental. Therefore many are supped - up to their
max. One supped - up program (faring very well as such) has been made to run very powerfully in tournamcnts is has competed in, the Sargon III. Commercial models are much less impressive.

Software vs dedicated machinesSubstantially stronger than the Sargon II, the III is the best software (diskette) program out. However, since the program was designed for the casual player already owning a home computer, the manufacturer cannot economically staff a team of programmers to approach the strength of state-of-the-art dedicated models. The weaker programming in conjunction with the limitations of home computers (previously discussed) in for instance a standard Apple
computer (running at 1.2 Mhz) will play a slower, weaker game.
Still, one can achieve a 1600 level game. The diskette is a good buy, but there is still some charm lost from having to key in moves, watch the board on a screen and then move the pieces on a real board. Playing against a dedicated machine that actually “thinks” is quite different. Many people owning a home computer will ask for a chess diskette simply because they already own a computer. What they mean to ask is - What will give me a good game of chess? (g/n)



Familiarity breeds contemptSome computer owners have tent their machines in for an update (eg. up-grade to a Budapest program) and were disappointed to find that the improvement was very slight, so slight for some that the machine actually seemed weaker. This may have been due to the human having improved against the first edition and improving in the couple months it takes to get it back. However, chances are that the player was comparing his cumulative win/loss ratio on the older version (where he may have had many more initial losses) to his current record against the update. Incidentally, the Glasgow program is a more significant upgrade. Although the top program out, the Super, boasts an official rating of expert, a number of the
stronger, seasoned chess computer vets on gut feeling estimate after a solid month with the machine, even its strength to be slightly under 2000, high 1900’s (without utilizing the programmable opening book).
These prosusually underate chess computers by far more thans 25 to 50 rating points. Maybe they need more time to attempt uncorking any quirks they may discover. These hard cores know exactly how to play against computers. The human player has a distinct advantage – he learns how to play against the machine, even subconsciously picking up on its weaknesses.

Two human opponents playing regularly against each other will attempt different lateopening/middle game lines and after the orther sees that he is losing, he adjusts and unleashes a different middlegame strategy. So, the ultimate goal is to produce a machine varied enough, good enough that you never catch on to his tricks; you can never learn to beat it automatically. Lacking that, a machine that learns to play against you. Well the Super has achieved the goal on both counts to some compromising degree. Opponents will find that they are actually improving, not just uncovering weaknesses in the machine. With the earlier, weaker models, once you discovered certain failings (entire books elaborate on these deficiencies), you could beat it at will.

Players would go through all kinds of contortions trying to make the machine’s game better, such as opening up files one usually wouldn’t, playing unsound sacrifices, or just playing an inferior game. With the Super’s advanced middle game and endgame positional understanding, its weaknesses are much more subtle (a master’s are even more subtle) and hardly as predictable. To most people the machine is unbeatable (excluding training levels of course). But for the expert or master, as they try different lines and find ways of outwitting it, i.e. learn the machine, it won’t quite learn back. However, with the owner’s assistance the computer can, in a sense, learn how to improve on its play and “learn” to beat its master. No long does familiarity with a computer necessarily breed comtempt. The computers now are so good that even after becoming intimately familiar with their style, their playing strength remains about the same.

Source: 11-1986, International Players Chess News: A History (Recap) of Chess Computers.
(Slightly edited by Hein Veldhuis.)

Last Updated on January 16, 2010


Um comentário:

Anônimo disse...

very interesting indeed, bravo