Is Fuego actually 4k?

I played a casual game against Fuego today - supposedly a 4k - and easily beat it without trying particularly hard.

I haven’t been able to play frequently for the past few months, but when I stopped I was hovering around 9k… I’m assuming I did not magically jump up 5 stones during my off time… but I would also assume the bots would be ranked accurately given the number of games they play each day…

Just curious if anyone has any insight - here is the game in question:

1 Like

You made similar mistake as I did, on a bigger scale than me.

I played several games with gnugo, and just found out yesterday that those “ranked” is equivalent of non-rated because gnugo rank is provisional. It did not count and had no impact on my rank.

The same here, Fuego does not have real rank at OGS, only provisional. That means that your past games with Fuego is very much equivalent of “non-rated”. We can not trust the 4k ranking either on Fuego because it does not have real rank at OGS

The bot ranks do move though. It’s just the human opponent that doesn’t move when playing against someone who is provisional. Feugo’s ranking is based on all of the rated games it has played in the past.

That is not real. When the human rank did not move and only bot moved. This was assuming human rank was “correct”. Wrong.

Most human players played with bots because they thought their rank is “wrong”, therefore by playing with bots to obtain correct rank.

When bots used wrong human ranks to get a “provisional rank” it is garbage rank, like “nonrated” already

Fuego and Gnugo caused more confusion here at OGS. The OGS should not mislead users on “ranked game” . They should be honest that “ranked games” with bots are not available, please play with non-rated game only.

It’s no difference if you play with a human who is provisional. Your score won’t move, and theirs will. It’s just that some of the bots are always provisional.

I think humans play bots often because they don’t want to be social, or they want to play instantly, or they want to learn without embarrassing themselves.

I would insist that when bots do not follow “5 game to be come real rank”, then the bots does not have “provisinal rank”, the bot has a measuring rank that measure those who played it

A real rank is not just measuring rank, a real rank will impact whole system of ranks. A loss means decrease, rank and increase for others.

The provisional rank for human is only for 5 first games, that is my understanding, correct?


for a human, only 5 games on provisional rank. 5 is very small, can be ignored on whole system. But bots have hundreds maybe thousands of games on “provisional rank”. Bots “provisional rank” is not true “provisional rank”. I called it “measuring rank”, a different category of beast that is hard to say what it is.

Per Sensei’s Library, Fuego “plays as a ranked robot on KGS with a 19x19 rating of 2 kyu (December 2008, 2.5 GHz, 8-core).” Since there’s about 2 stones difference between KGS and OGS, my expectation is that Fuego would have a rating of 4 kyu here on OGS. And it does. Perhaps the bigger question is how you played so well against it. :stuck_out_tongue:


That is the question indeed, Mark.

My current theory is either that Fuego’s servers were overloaded while I was playing, or that my body was briefly inhabited by the ghost of Shusaku :stuck_out_tongue_closed_eyes:


I’m ~5k here and ~2k here. I played a test game against Fuego:

I have a few things to say:
1.Fuego doesn’t understand opening. It plays too small and allows you to get out of opening with huge advantage.
2.It can’t see moyo. Build whatever you what, it will allow it.
3.During fighting it doesn’t play most severe variations, so you can get off with slack moves.
4.It likes to invade and just die.
5.I wasn’t sure whether it can see dead stones, so I waited for it to pass first and we played many moves after the game was over.

Most likely it is indeed overloaded with games. Or people just resign to not play after the game is over.

By the way, this bot has a lot of games, and most of them are paused. Maybe we should do a clean-up once in a while?

I’ll try to play handicapped games with it this evening.


That seems to fit what happened in my game - the invade and abandon strategy and weak fighting especially. Occasionally it would play a move that makes me think it is stronger than me, but other times I was able to get away with murder…

I’d be interested to hear the results of your handicap game.

Thanks for the report. :thumbsup:

I am 6kyu here and I just lost by 7 points.

I have to add that I usually do badly against bots compared with equally-ranked humans and I can imagine that I can learn to win against it, but still, 4kyu does not seem so far-fetched, since people who are bad at playing it might play more often to practice.

Edited to add:

I played a second game and won by 75 points. The big spread is due to several unmotivated auto-ataris by fuego and other very bad moves. It is possible that fuego simply plays very badly when he thinks that he is falling behind.


Well if you are curious about your strength, the simply play few more games with people :smiley: I don’t think showing the game around helps in any way.

What is interesting in this topic is that people do pause games against bots and abandon them. I haven’t played enough with bots, but is there any benefit from it? If I understand correctly that bots ranks are generally provisional, it doesn’t seems to make a lot sense. Regardless - it feels like the option of leaving bot hanging on a pause for longer time should be disabled. Would you keep a window with a bot open on your local machine, just to end the game sometime in future? Probably not. Why should you be allowed to waste someone elses resources then?

1 Like

I played 5 games. 2 and 3 stones I won. And I lost three times against 4 stones. Handicap is free placement, which is somewhat awkward.
I think I can win 4 stones if I get myself into mood. I lost mostly due to reading mistakes.

I would if I could! I used to play at least one game every day, but my schedule won’t allow it anymore.

The bot is convenient because I can play correspondence and not feel guilty about leaving it hanging for a week and then playing 20 moves lol

Interesting… that suggests it’s playing at a level somewhere around 7 or 8k then.

That would make my win make a lot more sense