DGT Centaur – the latest sensor board

by Frederic Friedel
1/19/2022 – Remember what "Sensory board chess computers" were like, four decades ago? Bulky, heavy, with complicated installation, and very expensive to buy. Today you get elegant, flat boards, which run on rechargeable batteries. And they are immediately ready to play. The one we are looking at today even adjusts its playing strength to match yours. It is a pleasure to operate.

ChessBase 17 - Mega package - Edition 2024 ChessBase 17 - Mega package - Edition 2024

It is the program of choice for anyone who loves the game and wants to know more about it. Start your personal success story with ChessBase and enjoy the game even more.

More...

Here are some of the earliest Mephisto sensor boards I owned and tested

For some pure nostalgia, take a look at the unboxing, setup and operation of the famous Mephisto Millennium. And this is the massive Fidelity sensor board I was given.

How things have changed! Recently, an ad for the DGT Centaur caught my eye. I have had a very long and friendly relationship with the company that supplies virtually all the sensor boards that are used in top-class tournaments. I told them I would like to experiment with their new chess playing sensor board, in a children's chess group that had formed in an elite school in Hamburg. The owner and CEO of DGT, Hans Pees, immediately offered to send me a set. And this is what I got: 

The Centaur is flat – just 1.2 cm thick, and light – 90 grams without the pieces. It makes a very sleek and elegant impression. Its operation could not be simpler: you set up the pieces and press the start button. No connection to the mains required. The board runs on a built-in battery that uses a standard USB charger. My set came pre-charged, and now, some weeks after receiving it, I have yet to connect it to USB (Hans, does it run forever?).

With the pieces in the initial position, the board knows you want to play a game. Enter a move, and it signals a reply, using discrete circular lights on the chessboard.

A small screen on the right displays the current status of the game. The schematics on the right (from the instruction manual) show the functions of the display and keys.

I grew quite fond of the operation of the Centaur. If you make a mistake, you can simply take back a couple of moves. The sensor board and understands what you want, and you can play an alternative. If you make an illegal move the "from" and "to" squares flash, telling you to take back the move and play something legal. To start a new game, just put the pieces back to the original squares. No key presses are required for all these operations.

There is one key you might want to occasionally use: the question mark button provides you with hints. It also displays how good or bad it thinks these moves are.

Another neat feature: you can simply switch off the board, using the on/off button, to adjourn a game. When you switch the set back on, it is ready to continue the game from where you left off.

Centaur has three playing modes: Friendly, Challenging and Expert. The default is Friendly, but you can change it at any time, even during a game.

"Friendly" gives you chance to win. As with Fritz 18, the Centaur will try to match your playing strength. If you lose a game it becomes weaker, if you win it will be stronger in the next game.

"Challenging" is more demanding. I think it sets itself to match your level of play, but will itself be slightly stronger, making good moves more often.

"Expert" is full strength, which means you have very little chance of winning. It clearly doesn't care how strong or weak you are.

Initially, in Friendly mode, I lost all games. Then suddenly I won one, much in the style of games against Fritz 18. And then another, and another. That was quite motivating. Just like with traditional chess programs, it has been decades since something like this was remotely possible. For amateurs, at least.

And that brings me to the only complaint I have with the DGT Centaur. I showed it to the chess class (eight-year-olds and younger). They were absolutely delighted, and played game after game. But they didn't have the ghost of a chance of actually winning any of them. Why?

Because chess beginners lose games by dropping pieces, not seeing that they can be captured. It is the last thing the Centaur itself would do. So the computer diligently takes the pieces that are en prise and in the end can do nothing but win the game.

After ten or more games, the rank amateurs gave up – they lost interest in playing against the machine.

So my request to DGT: get your programmers to install a "Beginner" mode, in which the board will actually drop pieces – just like the rank beginners in the chess group I was experimenting with constantly do. I will help you to devise algorithms to make plausible oversights by the computer. They should mimic the type of mistakes people who have just learnt the game tend to make. A Beginner mode will make the set available to a much larger, and very important, target group.

The CEO of DGT explained: 

"One thing perhaps to point out is that Centaur adapts itself to the player immediately during the game, but not so much from game to game. There is a very small adaptation from one game to the next, but that adaptation is not significant. And in any case it is reset as soon as Centaur is restarted or when another “level” is selected.

So basically if one player plays a game against Centaur it will adapt to the current player’s strength at once. If for the next game a different player plays against Centaur, it will adapt to this new player at once (and more or less completely forget about the previous game and the previous player).

Perhaps this is also a reason why absolute beginners find it quite challenging to win or draw their first game. But when they do draw or even win their first game, the delight is immense."

While absolute beginners might struggle against the Centaur, I must hasten to mention that the Centaur is perfectly suited to anyone who is approaching or has slightly exceeded Elo 1000. And of course also for players who are much stronger. That is my initial estimation. For anyone in these categories the board is loads of fun. Here's a video with full operational instructions.

You can get the Centaur from a DGT dealer, or outlets like Amazon. The price is $288.

Coming soon: review of a second modern sensor board that is similarly sleek and attractive.


Editor-in-Chief emeritus of the ChessBase News page. Studied Philosophy and Linguistics at the University of Hamburg and Oxford, graduating with a thesis on speech act theory and moral language. He started a university career but switched to science journalism, producing documentaries for German TV. In 1986 he co-founded ChessBase.

Discuss

Rules for reader comments

 
 

Not registered yet? Register

jenyes jenyes 1/20/2022 01:16
I think one of the drawbacks from this "era" was the relative ratings of the machines and the levels being set. Even the "rating" levels (1200, 1400, etc.) were way off. There was also the problem of predictability. I had one of the "stronger" Novag machines (Scorpio or Sapphire or something like that) with a rating of 2383. It took some time, but I was able to win a game with no take-backs on the "highest" (usually just the longest time control setting) level. Of course I wrote the game down. It was the White side of a Giucco Piano. Some time later, we revisited the same opening line and the machine lost again - with the exact same moves!!
Michael Jones Michael Jones 1/20/2022 04:06
Forgive my cynicism, but I had a sensory board chess computer in the 1990s which was not bulky or heavy, required no installation and was very reasonably priced. It had 40 or so levels, and the lowest ones were indeed those of a complete beginner, where it hung pieces, walked into forks etc. At around level 10 I was getting close to an even score against it; the top levels would probably have been a match for a strong club player, but I suspect any titled player would have beaten it. The best part of 30 years later, the Centaur has fewer features and costs several times as much. It's like reinventing the wheel and managing to create a less effective version that what was previously in use.
adbennet adbennet 1/19/2022 10:11
Actually I've been reading some other reviews online, and the Friendly and Challenging levels work just the way I described except: instead of A Little Weaker and A Little Stronger, they are About The Same As Me and Somewhat Stronger Than Me. And based on the reviews, the programmers do indeed need help with the algorithms for making human-like mistakes. One reviewer described the engine's adaptation as like stringing the player along by matching the most recent error.
adbennet adbennet 1/19/2022 07:58
Re "Complicated": it's one thing for adults, something else for children. If they can set up the pieces and just start playing (seems to be the case) then it's ideal. But "new game" on a laptop seems even easier.

From the manual: "Beware! Game information for the previous game is deleted as soon as you finish setting up the starting position for a new game; your previous game can no longer be viewed." That's poor design. Even my wristwatch will remember laps until memory is full, and then forget the oldest ones. Of course I can delete laps via the menu, or delete them all at once if I wish. As a teacher I wouldn't consider one of these in a classroom setting unless I could review (some of) the games afterwards. USB transfer would be a bonus. Even for home use, imagine the disappointment if someone adjourns a position and then someone else wants to play a new game. One or the other will not get what they hoped for. By the way, the manual doesn't say how many moves it can replay in the current game. There must be a limit.

From the article: "After ten or more games, the rank amateurs gave up – they lost interest in playing against the machine." To be fair, most rank amateurs also give up against humans fairly quickly. But I agree it should have a Beginner mode, and ideally it would "match" the human's blunders. In fact the _level_ setting should not be Beginner, Friendly, Challenging, Expert but Much Weaker Than Me, A Little Weaker, A Little Stronger Than Me, and Much Stronger. It should be simple to do with a computation of how inaccurate the human's moves are, and then pick second-rate moves from the engine's multi-PV to correspond to that. Of course this engine probably doesn't have anything as sophisticated as multi-PV, but it certainly could. For the player it could be fun to try to sandbag the engine by starting the game with some weak moves. Of course the programmer could also have some fun devising an algorithm against that strategy.
Werewolf Werewolf 1/19/2022 04:35
I own about a dozen Mephisto Chess computers and I definitely wouldn't describe them as "Bulky, heavy, with complicated installation".
The build quality of these machines was far superior to the modern machines, like "The King Performance".

Best chess computer ever? The TASC R30. Beats anything today.
1