Sorting Out Chess Endgames | Hacker News Reader

Sorting Out Chess Endgames

rjlipton.wordpress.com

29 pointswglb31 comments

(most) Chess engines without end-game database will play horrible endgame to the extend that a lowly IM can beat the strongest chess engine on the best consumer hardware. At least this was still true last time I was keeping tabs on Computer Chess scene. But with 4-5-6 pieces endgame its not only perfect but it also effects game quality in the middle-games, where most of the games are decided between human-comp games.

Nowadays, Computer Chess scene is not as exciting as it was 10 years ago (not because of Deep-blue). Now it has come to a stage that it is less about knowledge algorithm tweaking and more about, fine-tuning and narrowing down opening books to few sets where a particular engine is strong at and use extensive positional and book learning from previous games on those or similar lines of moves and win games.

Now when you (or GMs) play against a computer engine, you are not as much playing against an engine, but playing against a huge database that has been fine-tuned over 100,000s of games.

This is why no chess engine author will agree to an official tournament games against a top GM, without opening-book, EGTB, pos-learning, book-learning.

I played chess a lot when I was younger and enjoyed it a lot. Then I joined a chess club. Biggest mistake I ever made. Everybody was cramming openings instead of playing the game themselves.

It really was terribly boring. My strategy for dealing with this was to open with a really bad but non-standard move.

This led to an immediate abandoning of all studied openings by the opponent and usually that led to a relatively quick win because they spent so much time on openings that they forgot how to play. Funny that.

Of course, against a really good player that will just get you slaughtered but the people that rely on knowing tons of openings to gain a favourable position in the mid game will fail miserably.

I quit because of that, the fact that in order to be competitive I'd have to cram openings too instead of being able to simply play and hone your skill against the wits of the opponent in front of you, instead of the dead corpses of all the GMs before whose games they studied.

I really don't want to play Caro-Kann all over again, I'd like to play my own games.

For you or anyone else that dislikes the memorization parts of Chess, you may find Arimaa (http://arimaa.com/) interesting. Personally I also like it better than chess because of the somewhat more simple/elegant rules while still leaving a game that seems to have a large strategic depth.

Of course it also has a great hacker aspect of being very difficult for computers, but I think that's really only a rather marginal side issue. The game really is a fun game to actually play.

Sorry to hear you ran into that. It sounds like a terrible environment to try to learn chess in.

But you can get quite good without spending much time at all studying openings. I hate studying openings, and I'm definitely better than most of the people who cram openings. Learn how to play chess well -- what you wanted to do anyway -- and learn just enough of the opening to get to a playable middlegame position (playing less-aggressive openings makes this much easier, and allows you to figure it out based on general principles), and skill, not memorization, will make the difference.

(Just in case you, or anyone else reading, were thinking of taking it up again.)

P.S. I wouldn't want to play the Caro-Kann either. ;-)

I pretty much gave up on it, the only thing I still enjoy is teaching kids to play and one guy from Amsterdam I occasionally play, we're fairly evenly matched so it's almost always interesting.

I learned to play when I was about 6 years old, my neighbour was a very good player (I believe master class, but not sure, it is very long ago), he gave me Max Euwe's books and I dutifully worked through them, we would play for days on end, and I loved it, the guy must have had amazing patience (see http://news.ycombinator.com/item?id=1370499 ).

In school I was pretty much unbeatable until I met Arno Bezemer, a kid living about 4 houses down the road, he taught me a definite lesson in humility, I didn't win a single game for the first year and half and even when I did win it was usually only because he was distracted or had a really bad day, he taught me a tremendous amount about chess.

He enticed me to join the club and I lost almost all interest very shortly after that, I think I only played in that club for about 6 months, and it killed any interest I might have had in competitive chess.

Eventually Arno became the strongest player in Amsterdam, he's in the top 100 players for the Netherlands, ranks 3637 for the world.

But he's dedicated his life to it, I don't think I could do that.

http://ratings.fide.com/card.phtml?event=1001396

edit: corrected the author of the books I learned to play with.

Update, check out this kid: http://ratings.fide.com/card.phtml?event=24116068

#97 in the world, only 16 years old. That progress up the rankings chart is amazing.

So here is a question for HN. What is a good boardgame which has

(1) rich tactics (2) does not need "memorizing openings" type efforts to be good at or where such strategies don't work. (3) is easy to set up and play (4) has some online community

I don't pay chess these days for much the same reasons as jacquesm. Miniature wargames are fun but a bear to setup and tear down. Most popular boardgames aren't very rich tactically, mostly dependent on luck/die rolls etc.

I guess Go would be a good choice but I can't get anyone to play with here and/or teach me some interesting tactics :-|. I know the rules but don't quite "get" it yet.

http://en.wikipedia.org/wiki/Chess960

Chinese chess is fantastic. It's strategically rich, but much harder to screw yourself in the opening than with regular chess. It's also different enough to be delightful to learn for people who already enjoy chess. Two examples that really charmed me:

1. There is a jumping piece which attacks by leaping OVER pieces and landing on whichever piece is on the far side. It's possible to defend against this piece by moving other pieces into and out of the path between it and it's target, effectively redirecting the attack.

2. Kings cannot face each other on the board, or leave a very small 3x3 section of their territory. This complicates endgame, but makes your king more powerful in locking off the options of your opponent.

Arimaa is a game that you can play with a chess set. Its designed to be easy to learn but hard for computers to play. There is an unclaimed cash prize for the first bot to win the annual open tournament. Google 'arimaa' to find the rules.

Frustration with that was what led Bobby Fischer to propose Fischer Random Chess, which randomizes the starting positions of pieces (within some parameters) to make memorizing opening books infeasible. But it never really caught on: http://en.wikipedia.org/wiki/Fischer_random_chess

That's a very interesting solution to the problem. I wasn't aware of that, really neat. Thanks!

I vaguely recall reading about a computer chess tournament in the 70s where one program did not have an opening book and was optimized for making moves quickly. Its opponents would use their opening books to make lightening-fast moves early and then take much longer per move later in the game.

This program was beating much "better" programs because it would open in non-standard ways and while the other programs didn't play the non-standard openings particularly poorly, they weren't tuned to manage their time wisely and would use tons of time early in the game, then run into time trouble later.

I don't think that would work today any more, the amount of hardware that is thrown at this problem because of the publicity involved is immense, so a smaller contender would be outperformed even if they came up with a wild and unanticipated strategy.

I think it came as a surprise to everyone back then. Given limited storage, the opening dictionaries were probably restricted to the most likely opening sequences, especially when playing against other programs with opening dictionaries. Today you could probably just expand the dictionaries to include many of the unlikely opening sequences and couple that with a fine-tuning the investment in each move taking the clock into account.

"It is a bit funny to recall we paid millions of dollars for this amount of memory, when today you can buy it for tens of dollars—and then put it in your pocket."

I wonder how much research could easily be delayed a couple of years until it becomes cheaper to carry out. Was it so important to do that experiment back then?

"Hector, always logical, said he never used the machine, but he knew they ran “hot”—they were known to have cooling problems. So he kept his on all the time, and used it like a space heater."

Again, it seems as if responsible use of tax payer money was not really a concern at that place...

I wonder how much research could easily be delayed a couple of years until it becomes cheaper to carry out. Was it so important to do that experiment back then?

one of the goals of academic research is to take high risks (and costs) to explore possibilities for the future (looking forward 5, 10, even 20 years). if people only invested in 'affordable' research that had direct practical applications within a 1 or 2-year window, then the likelihood of 'game-changing' breakthroughs appearing would be much lower.

I have no problem with high-risk, long-range research projects as long as they are funded voluntarily and no fraud or misrepresentation is involved in getting the funding. You seem to be forgetting that the tax money is take by threat of force.

> You seem to be forgetting that the tax money is take by threat of force.

That's a fairly extremist position, there are lots of people that would voluntarily give a portion of their income up for taxes if it would lead to a better society.

I'm one of those, and I'm fairly happy to give up my 40+% in return for a relatively safe society, subsidized health care and reasonably good public roads and free education, as well as a catchnet for people that are down on their luck, even if I personally probably have a bit less expendable income because of that because I produce more that goes in to the system than I consume. But maybe next week I'll have a car accident and a half a million dollar bill will result and I will still have a house to go back to.

Now I realize that not all of that money is spent wisely, and maybe we could make things more efficient, but societies cost money to operate, and societies that are 'social' (as opposed to socialist, it seems to be a common mistake to think of them as identical) cost a bit more.

There must be some level at which you would object, no? On what grounds do you object when the percentage becomes 50%, 80% or even 100%? You've already conceded the principle and a slave is still a slave regardless if he cooperates with his master. Moreover, nothing I said implies that roads, education, medical services or even charity would not be funded -- in fact you proved my point, these things would be funded voluntarily precisely because many people (myself included) value them. Funding these things by force is immoral. If you think my position is extreme, try not paying your taxes for a few years and learn what the "threat of force" really means to your own freedom and right to property.

I don't think you understand what the word slavery entails.

I can stop working any time.

No master tells me what to do or when to do it.

There is no punishment for not being productive.

I am free to go where ever I want.

and so on.

Comparing taxes to slavery is pure hyperbole and whatever cause it is that you are promoting here you are harming it by association.

Aah, Churchill's old insult: "Madam, would you sleep with me for a million pounds?"

"Umm, well, I suppose so."

"Her's a tenner, lift your skirt."

"What do you take me for?"

"We have already settled the question of what you are, now we are haggling over the price."

Makes for a humorous story, but it's a terrible reduction ad absurdum argument. The fact that I am prepared to give up x percent of my income in exchange for a given set of benefits that have value to me says nothing about how I would feel giving up 2x or 3x percent of my income for some possibly different set of benefits.

Speaking as an Ontarian, I'd be horrified if a political offered to cut my taxes in half and give me the "freedom" to choose my own health care in the "free market."

That doesn't make me a slave, it makes me someone who pools certain funds in a coöperative manner. If you want the same argument back at you, try dismantling your army and giving every citizen the right to defend themselves from foreign aggression, buying their weaponry and training on the free market.

Choosing to voluntarily pool some portion of your income with other citizens to fund a common defense is not slavery either. It's another kind of freedom.

I think that was the unwanted machine.

How costly is it to use a million dollar computer as a space heater? Exactly as costly as using a ... space heater as a space heater, by the second law of thermodynamics. That also why all electric space heaters are essentially 100% efficient (thought energy is wasted in transmitting the electricity to the heater and that's lost in the form of heat too).

Isn't it very inefficient to heat with electricity to begin with? That's what I was told anyway. You are right, transforming electricity into heat is probably very efficient. But generating electricity is more inefficient than generating heat directly, hence heating by other means is a lot more efficient.

He might be excused, though, if it simply was another era when there wasn't yet any kind of consciousness for that kind of thing.

Hold a second here, when you're from Europe you may not know this, but in the United States and in Canada 'baseboard' electrical heating systems are very common and natural gas systems (like you see all over Europe) much less so.

Other heating systems common in the States are based on fuel oil ('red' diesel) that is atomized in a central burner and then used to either heat water (like in most European central heating systems) or air.

Elecrical forced air heating systems combined with air conditioning for the summer are also pretty common.

In Europe electrical heating is a rarity, unless for additional heat (small 1KW space heaters and such).

Thanks - I am from Europe. So why is electrical heating so common in the US? Or is it really a legacy from the days of SUVs and cheap energy?

Partly a legacy: 'all electric' was a selling point in the middle of the century when lots of new houses were being built. Partly cost: if you already have a modern electric hookup, the equipment is cheaper. Also, I think a lot of subdivisions and rural areas in the US were built without gas hookups.

Whether it's less efficient depends on what metric you're using. In areas that have hydroelectric or nuclear generation for electricity, the CO2 emissions are less with electric heat. Cost varies: there are places where electric simply costs less, particularly if you can spot heat instead of using a central system.

Here's a couple links:

http://www.iwilltry.org/b/projects/comparing-natural-gas-vs-... (concludes that switching to electric makes cost sense in BC, Canada)

http://www.leonardo-energy.org/webfm_send/377 (European view concluding that from a CO2 perspective electric is a win in Austria and France)

I think that varies by area. I've lived in Indiana, Texas, Georgia, and California, and I've had natural-gas heating in three of those (all but Georgia). I think pretty much every house in a populated area in the midwest has natural-gas heating, though I hear fuel oil is more common in the northeast, and electric heating is more common in areas where heating isn't very often necessary (like Georgia).

Yes, that makes sense. In Western Europe domestic heat is almost exclusively natural gas.

Hmm, interesting. My European relatives are all in Greece, where natural gas is still a fairly new thing, only available in major cities. For a while, electric was the only alternative to fuel oil, which many people preferred to avoid if they could (produces soot, requires you to get the oil delivered, etc.). Since there's a big difference in daytime/nighttime prices, the electric-heating units, unlike American ones, are big bulky things with a bunch of batteries, which they charge up at night at the cheap rates.

That's really interesting because charging and discharging a battery costs 8 eurocents per KWh in wear on the battery alone.

The difference would have to be substantial to make up for that.

My impression why large parts of the States like electric heat are closely related to your Greek relatives, America is huge compared with the number of people that live in the inner portion of the continent, so the only places where adding a natural gas delivery system would make sense would be the densely populated ones.

Then there are large areas that have only a short heating season, there is the convenience issue (very little maintenance compared to heating oil or natural gas), and the perceived risk (which I think is largely misplaced).

The recent changes in the electricity prices have made people more conscious about this and there seems to be a trend towards more natural gas for heating purposes.

Reply on news.ycombinator.com