Edit: Turns out this topic is about amending the TOS or something, hop in 🤷

I think you are misunderstanding or intentionally misrepresenting what I am saying. My ultimate point is that ranked games should be fair contests, which forbids the following:

  1. Either player intentionally throwing the game (e.g., in some styles of teaching games, the teacher tries to let the student win).
  2. Players offering each other assistance and advice.
  3. Setting a predetermined outcome (e.g., the teacher must win).

Thus, teaching games that involve the above are not allowed. Another form of “teaching game” that is essentially just playing a ranked game in the “normal way”, but then reviewing afterwards, is perfectly fine.

I think you are going about “teaching games” in a completely wrong way, especially as you are in the position (being 15kyu) to be offering teaching games to early beginners. As a teacher, winning against your student is not the point of a teaching game. The point is to teach, and with new players that are just beginning in the game, the bigger point is to not drive them away from the game by pointlessly flexing your skill over them. Focus on building their interest and curiosity in Go. Displaying your greater strength by beating them down is a needlessly vain activity that can be terribly detrimental in discouraging beginners. Keep in mind that the sort of players that are much weaker than you, which you would be teaching, might often be young children and could be particularly sensitive to such negative experiences.

I strongly encourage you to read this article: How to Teach Go | BenGoZen

To be frank, I’m engaging in the conversation less due to some idealistic concerns about protecting the ranking system, but more due to concerns how you (@espoojaram) might interact with other weaker beginners in misguided efforts of “teaching” that could potentially cause more harm than good.

I think the issue of whether or not beginners get an accurate rank quickly enough is a secondary concern to encouraging their participation and stoking their interest in the game. It’s perfectly fine if a beginner doesn’t play any ranked games at all, and just wants to learn and explore with unranked games, solving puzzles, etc.

I feel like whether or not a beginner has a satisfactory introduction to Go has less to do with whether they get accurately ranked on a Go server quickly enough (even still there are better ways to handle this, through potential adjustments to the ranking system rather than having them be beaten down in ranked games), but more with how their initial interactions with the Go community helps to facilitate their learning and introduction to the game. Expecting beginners to first need to lose a bunch of games against stronger teachers, in order to get their ranking adjusted, just seems incredibly counterproductive to the growth of the Go community.

3 Likes

image

Literally the rest of the points in the post seem fine without this, and I don’t want this discussion to revolve around who has what rating, who has progressed slowly or quickly, who’s improving or not at Go, in order to give their arguments or viewpoints credibility.

2 Likes

Not my intent at all, please re-read:

I was pointing out that the lack of volatility in his own profile could make it seem like the OGS rating system is more consistent than it actually is, leading to the belief that it assigns a set value that very accurately captures one’s “true” skill level. I did not mention a flat line or the actual rating (in fact, I noted improvement).

This was in order to give context for my following comments about the nature of ranking systems and how assuming them to be extremely accurate can be problematic.

3 Likes

I know, this one isn’t necessarily as negative as the one that came before it (not by you) but the trend is the same

I looked at your profile/rating/rating progression, saw that X seems to be the case about it and therefore Y.

It just seems unnecessary to bring that into it. If you remove that segment from your post (not saying you should) I can only believe it improves what you’re trying to say.

A number of people don’t like the rating system, they think it can be too volatile for one reason or other, they could call it a ‘mess’. You can find players and bots with a very volatile rating, typically from playing a very large number of games quickly or simultaneously, and one isn’t necessarily sure is it the system itself that isn’t able to handle such a thing, or is the level of play itself that’s the culprit and the system is just adapting to it as it would any other player.

Anyway we don’t need to bring their rating or their rank progression into it in order to explain why they probably have the viewpoint they do, if you see what I mean.

I haven’t really commented on this yet, but my issue might be similar to @ArsenLapin1’s, in that the logic of it seems a bit backward.

Rather than saying

you can play the ranked game how you want to, but if you’re the stronger player and you don’t win we need to annul it

we should just be saying

If you’re not playing the game in the way that suits it being a ranked game, then don’t play a ranked game, play unranked.

That’s not really about whether it messes up the rating system per say as a whole, it’s not about whether you should’ve gained or lost a point or two here or there per game. It’s just more of a general principle where you shouldn’t do things that intentionally manipulate your rating, because the general population find that annoying.

People want to play rated games against people their own rank, or against the rank they see when they challenge/ accept the challenge. They don’t want someone who just resigned 50 games to drop down 10 ranks to crush weaker players, they don’t want someone intentionally rating themselves lower to beat higher players for some kind of fun, and to take their rating points away unfairly. There’s lots of situations like this, and it’s not about the current rating system implementation, be it glicko, glicko2 or Elo or anything else, it’s more just on principle I would say.

2 Likes

I understand what you’re saying, though I would argue that it wasn’t unnecessary – some example is needed to give the context that large fluctuations are the norm rather than the exception, since so many conflicts in earlier discussion occurred because of the assumption that an assigned rank is stable. I can use a different profile with a very smooth ranking line if that helps, but I think context from one’s own experience is often better than unfamiliar examples.

I also fully agree that the rating system may seem like a mess, and that people are completely justified not to like it. I just wanted to show that other points of strife (trying out new styles, playing the occasional suboptimal move etc.) are generally insignificant contributors to that volatility compared to the general mess that exists inherently.

Maybe a better way would be to link to some random profiles that fluctuate a lot and just say “this is the norm”?

1 Like

I think I understand what you’re saying also. I guess I just took some exception to the phrasing.

In a vacuum I guess I agree, but we already have one player either leaving or wanting to leave this discussion when their rating and progression was brought into it, so I guess I’m being overly careful in pointing it out.

Yeah I mean possibly - even screenshotting some rating graphs rather than linking a profile could be even better. I just wonder if I was the average anonymous user, and found that my profile was being used in a random discussion, how would I feel :slight_smile:

1 Like

Agreed, and I acknowledge that my cause-and-effect assumption (that a user’s experience was in fact the source of a misconception) could be inaccurate. Amended way of getting the point across:


This type of rank graph is actually less common, and usually indicates a smaller amount of ranked games.

Volatile rank
This type of graph is very common, and doesn’t usually occur because of someone not taking the ranking seriously enough. Instead, it occurs due to natural variations in playing style, mood, luck, etc.

If we wanted to produce the first graph more often, we could use a different ranking style (like some other servers) with “inertia” that requires certain events to rank up or down. That type of ranking comes with its own issues and is a whole other can of worms.

1 Like

There are just too few games. More games are needed to see waves. So its norm.

Yes I definitely realized that, but still wanted to provide the context that – in general – more variability happens. Pretend I rephrased it to “this is not the norm for an established account with a lot of ranked games”. Sorry this caused so much trouble.

1 Like

I wouldn’t say it/you caused trouble, I think it’s an interesting addition to the discussion overall, I was just commenting on a trend I was seeing in the thread.

2 Likes

You’re good, I don’t think referencing profiles has ever really been a point of contempt in the past. Just happened that someone else had written almost the exact same pattern of words earlier, but instead the subtext was “you haven’t improved in years” instead of “your improvement has been linear” :yum:

2 Likes

Bringing my rank into consideration was utterly logical, and also remarkably elegant writing, since it builds up into the following arguments.

Most of those arguments, in my perception, are unfortunately illogical or misguided.

So much so, in fact, that I don’t think I possess the ability to respond to them in a way that is both convincing and non inflammatory, which is why I’ve decided to “hemlock” myself.

(I guess I could just stop replying, but I’m guilty of being arrogant and proud and I can’t stomach y’all suspecting I’m running away unable to admit I was wrong or something)

It’s a difficult decision for me, especially because the person making those arguments that I perceive as being illogical is a contributor to the code, which means that if I managed to convince them, I’d be in the rare (for me) position of actually being able to make a positive change to OGS, for once.

But without the ability to make non-inflammatory arguments, it’s pointless.

Assuming you’re talking about me (given that I am the code contributor who mentioned something tangential to your rank):

I wasn’t bringing your rank into consideration at all. My opinion and logic stands regardless of your rank. I wanted to bring into consideration the apparent smoothness (in a mathematical sense) of your rank graph, because I thought it might be leading to a misconception about how stable ranks on OGS are in general. However, I realize that I may be completely incorrect about that assumption, and you are more than welcome to correct me.

I am fully willing to admit I am wrong, and will not take things in an inflammatory manner (so long as they are not outright insulting).

Then we both want the same thing, and I am happy to hear your critique.

If I say I don’t treat someone telling me I’m wrong or that my logic is flawed as inflammatory (as is the case), are you able to make non-inflammatory arguments?

2 Likes

I appreciate that – I wasn’t worried that you might take unwarranted offense, though, I’m worried that my response will be objectively inflammatory.

There are few things that make me as bitter and emotional as having to make extensive effort to “debunk” arguments that I perceive as being illogical, and that’s likely to cloud my judgement and make me veer into rhetorical fallacies and a smug tone without realizing in the moment.

I might try to write a response over time, hoping that re-reading what I wrote after some time will help me keep things civil.

In the meantime, you have brought up the idea that the volatility of the ranking may reflect actual fluctuations in the player’s ability, and thus momentarily make the matchmaking more accurate.

This is an idea that other people, including myself, have proposed in the recent past, but since it’s just a hypothesis, we’ve been discussing the possibility of testing it to see if the evidence backs it up.

(And after thinking about it quite a bit, I believe it to be wrong)

While I have many ideas in merit, the main thing I lack is the general programming ability and the specific understanding of OGS’s code that would be needed to put it into practice, and the ability to autonomously gain those abilities; so unless I receive help, it’s unlikely to ever happen.

So if you think this would be worthwhile, I invite you to read this thread and consider helping us (me?) out:


By the way, when you brought up my rank you said

So while you then used the atypical smoothness of my rank to develop your further arguments, you did have a good reason to talk about my rank in particular, conjecturing an explanation for me apparently idealizing the ranking system. It makes perfect sense.

I mean, you were wrong in this case, as my understanding (whether right or wrong) of the ranking system comes from having thought about it a lot and having looked at a lot of other evidence, but your inductive reasoning made a lot of sense nonetheless.

Fair enough, and I appreciate that you recognize this.

I don’t think I have brought up this idea, at least not intentionally. What I brought up was nearly the opposite: the idea that there is really no such thing as a true/ideal/correct rank or measure of a player’s ability, and that volatility in ranking represents (in part) this underlying fact and not any momentary variation in playing ability. In the thread you linked, I’m arguing against the “cultural perception that a player’s ‘true strength’ is pretty stable over time”. I’m saying that the informal perception of a stable “true strength” is fine, but most attempts to capture that formally and mathematically will be flawed because the concept of a “true strength” is flawed, or at least needs a relatively wide margin to be in any way accurate as a predictive measure.

For example, take some imaginary ranking system and consider one player who may be really good at reading and higher ranked than another who is really good at positional judgement. Maybe they play each other, and the weaker player ends up winning because their fuseki happened to give an advantage that reading couldn’t overcome. Maybe they play a few times and the same outcome (the “weaker” player winning) happens each time. However, the “stronger” player still regularly beats other players at the same rank as that weaker player (under this hypothetical ranking system), just not that specific player. Which player’s “true” strength is higher? Was this a violation of the ranking system?

(I do realize that a ranking system is generally probabilistic model, so the higher ranked player in this example is higher ranked because they win a certain percent of games against a certain rank on average, but I hope this example shows how various factors can cause problems when a rating system continuously updates with every new data point)

This may seem contrived, but this kind of mismatch of skill areas happens all the time in ranked games. It gets even worse when you consider the fact that a player may stumble upon moves that their opponent is ill-adapted to handle, not out of skill, but out of pure luck. Even though Go is a highly strategy-based game, human nature is still a confounding factor. That’s part of what makes the game fun.

Glad to have that cleared up – hopefully further discussions of rank should be easier, and I apologize for my mistaken assumption.

If it helps, a lot of my comment was building up to point out what I saw as a problem in this sequence of comments:

The main thing I wanted to convey after looking at this conversation was that ArsenLapin’s comment describes a pretty standard outlook and approach to Go, and doesn’t affect the ranking system nearly as much as ranked teaching games or general volatility would. And, even if it did affect it to some degree, it wouldn’t matter, since the ranking system is mostly a matchmaking tool.

As a side note, looking back I must say I’m surprised that it was my comment on the properties of a rank graph that caused so much potential for inflammation, and leads me to believe it may have been lumped in with the comment next to mine which actually mentioned a player’s rank and used it as justification for a number of statements. So, just to be sure, the earlier comment about lower ranks failing to execute a proper teaching game (and using them as a form of vanity) was not mine.

I also realize that much of my explanation is in the context of the current, volatile system and that other systems may handle such volatility more gracefully. However, I still stand by the idea that a “true” rank is elusive, and any system that attempts to capture one will be subject to the idiosyncrasies of human nature in some way or another. Often, perceived volatility is a product of these systems attempting to compensate for human nature.

It’s been interesting exploring this aspect of the argument but I think that the core point is not about whether a person’s true rank is represented, but rather “the amount by which beginner’s ranks are wrong”.

The former is elusive. The latter is categorically the case.

Agreed. Maybe we can continue the volatility discussion on the other thread? I’ve posted some more there: Testing the Volatility: Summary - #10 by joooom.

Oh hey, it’s me again.

Just wanted to update that some time ago I did read this blog post that @yebellz brought to the table, and I completely agree with the philosophy of it – which I already did, actually. But thanks for bringing it to my attention, I’ll try to apply the advice as much as I can in my future teaching games :+1:

Yep that blog has inspired many.
I remember an article about teaching beginners, in a AGA yearbook published before 2008. An excellent article too if you can find it.

1 Like