A Newbie drew Stockfish at highest level (2 draws so far)

Hi people,

I'm trying to beat Stockfish 10 in my smartphone, using Droidfish app.

I'm doing it for the sake of human kind, since lately GM don't play serious official machine-human games to restore human honor.

I'm trying to prove that machines can be faster than humans in chess, but faster doesn't mean better. But the problem is that I'm an average chess player, maybe we could say a newbie. So, in order to success I'm using mental brute force (no cheating).

The Method:

What I'm doing before each move is to think ahead at least 5 posible lines, visualizing 10 moves (20 half-moves) deep in each variation. If stockfish in my phone can calculate that deep, I have posibilities.

The Rules:

The rules are that I don't touch the pieces while I'm thinking, and I can't look or analyze using an engine or any other sources. Everything has to come from my own sweat.

But, since I want to prove that machines are faster, but not better, I'm allowed to leave the match and go to work, or go for a walk, etc. and keep on later or the next day. I'm allowed to have a life beyond the board.

This rule is obvious, but I'll mention it: I can't take back a move. If I make a mistake I'll keep on with it.

Hardware:

My smartphone is an LG X power2 (M320N). It is calculating 1.7 million nodes per second.

My body is just a normal body, powerd by food and air, and some cold drink from time to time.

Time control:

In order to get quality moves from the machine, I'm allowing my smartphone to think for at least 40 seconds before each move: that's 11 million of positions.

For me, I usually take from 5 to 15 minutes to think the five or more variations, 20 moves deep each one, in middle game or opening.

Progress:

So far I have being able to beat stockfish at levels up to 36% (level 7). And I have being able to draw at levels up to 72% (level 14). So, now I'm trying to beat it at level 99.9% (level 20, hightest level) and I almost draw , so I can see there is a glimpse of possibility.

I think nobody before made this kind of experiment, thinking that way, with this time controls, etc... so I'll keep trying.

I'm posting the game in which I almost draw at level 20. I can post on request the rest that I mentioned, but I just didn't like the idea of filling the forum with games that are not so interesting.

------------------------------------

I would like to know your thoughts about this, if my smartphone is too much understrength and I should give it more time to think... if there is someone I didn't know that did this before... if you would like to do the same experiment, etc. All related stuff is wellcomed

(PD: My user name in my phone is Arahat)

You were dead lost at move 18. You just didn't know it. It is a common beginner’s mistake to equate length of game with degree of struggle.

Yes, it is likely 0-1. 

I checked with SF. White cant make a fortress.

That's quite possible. The endgame is the most difficult phase when playing against an engine, because checking their tables they know every single endgame position.

I'm working at improving my endgame with "Dvoretsky's Engame Manual" book. For now I only had time to read the blue print. I still have a long path ahead in endgames strategy, since I just finished the blue print

For a newbie beating Stockfish is like 1 in million chance. Even for top 2700  GM , the chance of beating SF is like 1:100. Simple enough, that is why people dont try. 

P.S. in my young age, about 20 years ago, it was possible for me beating older version of chessmaster a couple of times in those old hardwares.

According to the analysis I figured out that in move 25 I should have gone to the g-file to stop the black king supporting the pawn promotion.

Then, after stopping the promotion I think it could have being good to try to get a passed pawn with a breakthrough in the queenside, but only if I get my pawns more advenced, because I didn't have pawn majority in that side.

Becasue I didn't have pawn majority in that side two things could have happened: pawns would have being blocked each other if I'm not able to perform the breakthrough correctly, or if both of us get a passed pawn it would have become into a pawn rush to get a queen, in which case I could be screwd if both of us queen at the same time, because his king would have being more protected than mine after that promotion...

But if you couldn't find a fortress for white I would have lost anyways, because that was my other plan after the pawns being blocked in the queen-side.

Let's see how next game develops... I'll play black in the next game, but I will try to think even more!

How strong is today's Stockfish?

Stockfish is extremely strong today. 0.1 sec per move thinking of SF is like 3000 rating and 1 sec per move is 3200 rating.

In comparision with Deep blue, SF on smartphone is like 3500+ rating and can beat Deep blue in at least 99% scores.

Someone has to try to beat the unbeatable, for gods sake!    I can't stand a machine being better than humans in chess.

One day, maybe in 2 years trying I'll win that match. Maybe sooner, who knows... 

I think that in my phone stockfish is not so strong, and the way I'm thinking so much can give me a chance. In a formal two-hours match I wouldn't even try.

If in the end I don't win, it would have been a good training anyways, and I like the process, since I like looking for good moves ahead, it makes me feel relaxed and forget about problems (those that don't have a solution).

I'm allowed to pee, and I'm allowed to order a pizza. But not at the same time, that would alter the challenge if the phone falls into the toilet.

Fair enough,

1. If you have the chance to live 1, 000, 000 ( 1 million days) and play one game per day.

2. And still using today's stockfish for the future, without any update. 

3. Then, you can beat Stockfish before the year 4700.

Very Good idea. Creative. I wish you luck to beat Stockfish, you will need it. Previous game were interesting and you shouldn't give a pawn somewhere in the middle of game. In endgame you were 100% lost, but you aren't too far from draw or even a win. I support you maximally. And, once more, Good Luck!

Chessmaster's top level on my old RAZR was pretty easy, and I beat CM 2100 once or twice on my desktop. But even Fritz 7 was too much for anyone but top GMs. Stockfish is untouchable unless you're Hikaru Nakamura.

You're not gonna beat Stockfish, but what you wrote here indicates that you are learning something of value, so I think your exercise will produce rewards that justify the effort.

Thank you very much. That boosted my morale.

That's right, one single pawn is decesive. Next time I'll try to keep it.

The truth is that I'm playing thinking that I'm going to lose, so I was content with a pawn loss thinking it could have being worst.

Next time I see a line in which I lose the pawn, I will keep searching believing that I'll find something better

Thank you all.

What keeps me trying is that I think this is a good training in case I don't win. I don't feel like wasting my time.

They say that playing long timed matches with superior opponents improve your chess. Coupled with studing some middle game theory books and the endgame book I mentioned... (whenever I have time, because some days I don't have time). I think one day I will improve enough to beat stockfish the way I'm playing.

That match in which stockfish only won by two pawns gives me hope. Let's see what happens the next match.

I know, it looks impossible. Something like winning the lottery. But I strongly believe I can do it before the year 4700. Because of the method I'm using. I think nobody ever before did it this way: in previous matches GM vs Machine they always play a formal 5 hours match. But even though machines are faster, does doesn't make them better. We can think the same positions they think, it just takes longer, that's all. So I think I can do it

Ah yeah, I had a coversation with Pfren about this a while back, the difference was that SF would run on a laptop and it'd run during all the time you think, in short, a daily game against SF.