Cheating with the score estimator

I actually made that argument, facetiously of course, in a Forum thread a couple years ago, but no one got it. Indeed, as I recall, it sparked outrage.

3 Likes

I’m out of popcorn, but I want that link.

3 Likes

I think it’s a strong indicator that analysis board is much closer to being able to see the board instead of playing go-the-mathematical-abstraction in terms of external help. But while I do find definitions of go such as Tromp-Taylor beautiful, I think using a board and stones, or simulating such on a computer, is very much within the bounds of legitimacy.

2 Likes

Even if you can explore a bit more variations and a bit deeper (maybe?) the DDK main problem will still be to select which move he has to explore, as exhaustivity is not really the way it works.

2 Likes

Or, even if you know what the score is and how areas are divided, the problem will still be to decide what to do with that information. Like, actually play.

Score estimate is a snapshot of the board, nothing more, nothing less. It’s not playing. I repeat: it is not playing. Analysis is playing; it’s putting stones on the board and playing out outcomes. Theoretically, someone could just play out every possible move under the sun and choose the best one, but that is not score estimate, that is using AI (let’s not hide behind our pinkies here).

You want to call people cheaters, go off; I get to call purists pretentious as well.

1 Like

Thanks for spurring me to look up the link. I don’t usually bother, as I don’t consider myself a research bureau for debaters. Except when I am moderating, I write posts from memory (sometimes inaccurate, I admit), and people can take it or leave it, it makes no difference to me.

However this is different, and fortunately, it wasn’t as hard to find as I had thought. This is the link: Game analysis...a crutch?. I have three posts there, which must all be read to get the full effect. Rereading it, I remain very happy with it as a piece of writing and still stand firm in the philosophy expressed.

3 Likes

It’s played. A good SE will play (good) moves (in the right order) to determine the score. Don’t think a score just as a set of points, it’s a result of reading.

Analysing is just a help to see more but that’s quite a small help. Thinking is still on your side. Approaching go by exhaustivity has always been a failure (see old go softwares).

2 Likes

I don’t think it works the way you think it works. During the game, the SE doesn’t show you moves that lead up to the score, it’ just shows you the score. So, even if it’s doing the reading for itself (which I kinda doubt), as you say, it certainly doesn’t share.

Also, I’m done with this topic, be judgmental all you want, it’s still not cheating.

1 Like

You are right, but there’s more: all of us are just able to think moves of our own level. I can’t imagine tricks and smart moves above my rank.
Likewise I can’t spot weaknesses that I don’t know how to exploit.
So when I use analysis I can’t actually be stronger than my rank. The valuable help is just to check more moves and variations than I could “read” in my mind only. That’s a lot anyway.

Yes, it does. That’s how it states which group is alive and which is dead.

1 Like

Well, I don’t think a rank is so ironclad that it can’t sway with a few moves read ahead here and there.

My point was, if SE-just a measly “this is the score and the areas as of now, give or take” can leap a player from DDK to Dan status (which I find ridiculous, sorry), then what analysis-“let’s clearly see how all the ideas I can think of play out” can do?

Consider near the end of the game, SE tell you you need 2 more points to win. This you will not know without SE and this is a Dan level information. You can use it as to try to get this 2 missing points, taking a bit more risks for example.
Another example: SE tells you the status of this group is not clear. Nice, another Dan level info that you can use at your level. Let be sure it lives (or let’s kill it)

With your analysis tool, if you have no idea about eyes, cut and connect then push from behind, shapes,aji,miai, then moyo/territory lines etc etc… The improvement will be small, neglectible.

2 Likes

I can tell you my experience.
When asking review for a game, a stronger player told me: “hey, look, if I play here your group is dead!”.
I didn’t see that move nor did my opponent. But that move would have changed the outcome of the game.

I don’t like how all this “cheating” thing came out I think it’s good for fighting, not for understanding each other.

I think that someone above wanted to say that SE could tell you that a group is dead when you didn’t notice. So, actually, it’s some piece of advice from higher level than yours. While analysis isn’t because we don’t actually try all possible moves with their consequences: bot do that, not humans. Humans only try moves that they think are reasonable… which leads us to what I said before: a stronger player knows shapes and moves and weaknesses that I don’t know, so I’ll never try them in analysis mode.

I don’t think anyone here is a cheater. I don’t like very much SE but still use it sometimes. And I use analysis sometimes. I do think they both are crutches and, well, I use crutches if I want to. :grin:
If someone doesn’t want them, there are triggers.

I used to disable analysis for myself, when I wanted to help myself avoiding it. I could do it again, if I want.
And if someone wants a “fair ground” were everybody play without SE and analysis, they can make their own challenges and tournaments disabling them. It’s fair enough for me.

1 Like

I think much of this discussion has become unproductive, since it was initiated and revolved around the incendiary characterization of this tool as “cheating”. It’s an unfortunate word choice, which has framed a lot of the discussion around personal attacks.

I think, by definition and convention, any use of the SE is not “cheating”, since it is a feature built into the site. If it were against the rules to use it, it would simply not be offered as a feature, but that’s clearly not the case, since it has been part of the site for years. However, I think it is valid to question and debate whether it should be available as a feature.

Ok, so here’s my elitist, condescending opinion on the matter…

Ultimately, to me, the issue of whether the SE (in its current form) influences games is largely theoretical and a matter of principle, rather than a major practical concern. Frankly, I don’t care if other people want to use it, and I completely support maintaining the current status quo. In fact, the only thing that I would strongly advocate for is to not upgrade the SE to perform better (e.g., by applying strong AI).

My position is precisely because the current SE is terrible, and offers little useful advice to SDK-level players and stronger, and I’m not so sure if it’s actually very helpful for DDK ranks either. I don’t think any SDK players take the SE advice seriously, and would only look at it for entertainment, rather than for useful advice.

However, imagine what could happen if the SE was upgraded to use a strong engine like KataGo or LeelaZero. The SE will evaluate not only the current position, but also hypothetical positions that one can setup with variations. Thus, a player could use SE to judge which of a set of potential moves is best, or to check if a tenuki is safe. Even if checking only the current position, a sharp change in the estimated score could provide a helpful hint that the opponent has left a weakness that should be exploited. Of course, with the current SE, it’s too atrocious to actually provide such helpful advice, but if it were to be upgraded with a strong AI, then it could be reliable enough to offer advice.

2 Likes

and how can you make these happen with such confidence that only SE stands in the way if you

That’s right, you can’t.

What about players who simply want a more accurate estimation?

I’m against upgrading the SE to give more accurate estimations than it currently does. Note: this is only in regards to the SE available during the ongoing game. I think it’s fine to have strong AI tools available after the game is over and only during the scoring phase.

2 Likes

Not really, maybe you didn’t notice because you just didn’t notice. It’s not “dan level”. Isn’t there a video somewhere with Cho Chikun where his opponent places a stone in atari or kills one of their own group, something like that? Did that person become DDK in a ripple in time or something? No, they just happened to not notice.

For this to happen it needs from the player to play a move and use SE before and after. To which I don’t really agree as a practice, but I won’t call it cheating because it’s allowed. This is not using SE, it’s using SE +Analysis, which is a different thing.

Theoretically, though, you could. The same way, theoretically someone could use SE with nefarious purposes or to make it do things that it’s not supposed to do.

Honestly, I think things are more fluid than this. Sometimes we think beyond our means, and usually that’s how we improve. Furthermore, I’d argue that analysis, where you are able to see up to move 5 (for example) with stones on the board, will “lead” you to think of move 6, because the rest of the stones make it make sense, while you’d never think of it just in your head.

I’ll just play devil’s advocate, but what if a player misjudged and passed, and then decides to resume from scoring phase because they thought they were winning but weren’t and one more move could make or break their group?

2 Likes

I think that’s a very valid concern, and it was brought up when I suggested to Use KataGo for Scoring.

However, I think that ultimately, we need to strike a balance between user convenience vs competition “purity”. It’s not ideal for what you described to happen, but at least it requires both players to make the mistake of passing too early, rather than being a tool available during the entire game.

2 Likes

I will go further and say that most weak DDKs and TPKs don’t use the SE at all. In my moderating experience, the majority of them do not count and go to scoring to find out what the score is. If they see they are close, they take the game out of scoring and typically make doomed attempts to improve their score. If they used SE, they would do this beforehand and not wait for the scoring phase. As for SDKs, I think the usual time controls make strategic use of the SE impractical in most cases of Live play. However, some might use it to get a quick score when deciding whether they should resign. “Oh, I’m 30 points behind with no prospects, yeah, I guess I should resign.”

6 Likes