The case against using computer chess engines during games on stream.

Hi there,

I've been spectating a bunch of chess streams on Chess.com's ChessTV platform, the St.Louis Chess Club's stream and others. One noticeable characteristic I've seen on streams can be described in the image below. 

The bar on the left side, that is highlighted, is an engine evaluation meter. Whenever each of the player moves, the bar may change by going up or down.

The more white the bar is, the worse black's position is. The more black the bar is, the worse white's position is.  The position shown above is slightly better for black, according to chess.com's engine.

Now, given we have a little bit of an understanding of what is being discussed here, I would like to make a plea for chess.com and all streams to not include the engine evaluation meter while discussing live games on stream. 

There are a few issues I find with having it there and I discuss them below.

1. Engines devalue in-game commentary: 

I believe this is the strongest argument that can be made against having the sidebar. Let's run through an example to help describe what I'm getting at here:

In the above screenshot of the game, Carlsen vs Grischuk, IM Daniel Resnsch & GM Robert Hess commentate in entertaining style as usual. They give their thoughts on various aspects of a position such as positional planning, strategy, discussion of sacrifices, etc. Their commentary is why chess.com's stream is as successful as it is. The way they talk and behave is why they draw in viewers. 

However, when include the engine bar, something noticeable changes about the stream's quality. The engine devalues a chess game's suspense. If I know the exact evaluation given by a authority without explanation, that leaves me always thinking: "What is the point of the commentary if I know the evaluation of the current position?" or "Why aren't the commentators talking about the engine moves (the correct moves)?"

It also leaves me with an empty feeling that prevents me from developing my own ideas about a position. If I know the evaluation of a position, that's often a sign whether various sacrifices and tactics work or are present in positions. I won't otherwise think for myself if I have the engine deciding what I should think about or how I should approach a position.

This is similar to cheating on an exam where you look at your neighbors'. If I see what my neighbor (assuming they are competent) is putting as their answer, that gives me an idea of what the answer is or could be. Consequently, that will devalue the test and your learning experience as a whole since you're spoon fed an answer. 

2. It's distracting:

Over 15,000-100,000 years ago, humans transitioned away from hunter-gatherer societies and moved towards ancient agricultural civilizations such as the Babylonian empire & Indus Valley civilization.  What were humans doing before this time however? Well, they were living on the savannah hunting and gathering for food and resources. Humans also have always been on edge given the numerous dangerous creatures and environmental factors present.

In order to defend ourselves from long ago, we had to constantly be on guard from dangerous animals such as saber tooth tigers, bears, mammoths, large rodents, etc. We had to evolve many traits in order to cope and survive in that brutal age. One of the traits, that was developed over time, to help us survive was our peripheral vision. It was useful in avoiding getting eaten, stabbed, or being hit by various objects. 

Modern day humans, however, do not have to concern themselves with dangerous animals for the most part. We still do have our peripheral vision though as a result of it being drummed into us through evolutionary processes. Whenever we see a moving object such as a small fly, an eye floater, or a bird outside of our window, we are drawn to the movement of these objects. They distract us momentarily from our daily tasks such as walking around or working on our computers.

The same thing happens while we are watching chess streams with an evaluation gauge. We are distracted and drawn away from the chess board. The commentary itself is devalued therefore as well by having to have to constantly look away towards evaluation gauge. Whenever it moves, I have to look at it. It draws my eyes away, out of instinct, from the game for a few seconds at a time. That may not sound like much in itself, but over time if you look 100-200 times an hour at the bar, that accumulates in time spent at looking at the bar instead of the game. It takes away viewing time from the stream.

3. You don't learn anything from the engine:

This is the most obvious one in my view. What are we learning on stream with the evaluation gauge? We understand that the position has an evaluation, but that evaluation is meaningless without context and commentary given beforehand. 

Commentators are often not even in sync with what engines say about a position and even if they are, their commentary isn't as creative or enhanced given the evaluation meter. 

4. It kills the suspense:

From an entertainment perspective, if I know what the ending to a game or movie will be, that will dull my viewing experience. What fun is life, if you knew the future and everything to come? You would be bored if you knew exactly what would happen at every point in your life.

The same goes for the evaluation meter: if I know the evaluation of a position, that takes away from the suspense and tension in a position. Games are more boring if you know what is happening exactly on a deterministic level.

Solution: 

I actually like how the St.Louis Chess club does their streams regarding the engine sidebars. 

Yasser Seirawan (on the right) and Jennifer Shahade (on the left) first give their thoughts on the position without the use of engines and analyze it until they are content.  Thereby, the audience is also usually satisfied with their analysis.

After about 10-15 minutes of commenting, their co-commentator and assistant, Maurice Ashley, brings up the position with various lines and ideas proposed by the engine. 

This helps bring new and fresh ideas that the Yasser and Jennifer haven't looked at and a conversation of ideas is born about how we can understand a chess position. 

I realize there is more nuance to be discussed regarding this topic in the future, but I hadn't seen this being discussed nor articulated anywhere before. 

I hope this post helps make the argument towards constructively using a chess engine and learning chess!

Sincerely, 

InfiniteFlash

Yes, STL style is a way to do it. All you need is a billionaire octogenarian who has more money than he knows what to do with.

Engine eval is useful for letting casual observes know what's going on. It adds to the drama and appeal, not detracts.

In this year's WCC match, I watched some 14 year old proudly declare the game will end in a draw (after only 15 or so moves) due to his phone's eval.

Sure there are idiots like this, but I don't think removing engines is the solution. The solution is educating players that practical play (even at the WCC level) is very different from objective engine-level play... that's often what the commentators do. They inform the audience that sure, it's 0.00, but black is in a hell of a lot of trouble.

If viewers can't understand this, then maybe they're 14, and there's not much you can do to fix that.

I’d love to remove it. We get no comments about the position just ‘0.00 easily drawn endgame’ Hikaru resigns 10 moves later ‘what an idiot, easy draw’.

I don’t even care what the best move is if nobody can explain why. It does devalue.

And I get that those people wouldn’t contribute usefully anyway but the engine muppets drown out anything that may be useful.

Good post by the way. There aren’t many on here.

OP, I dont know why you say engines teach nothing. It would be due to that you dont know how to learn from engines.

" For example, I played a few dozen games against stockfish with stonewall attck, bird opening etc. Stockfish stopped my attack easily. I learned how Stockfish stopped my attack."

Engines are the best tools for many things,

For example. . Compare your strategies and tactics( You may damage your opponent's king side attack in 6 moves, but stockfish may show you much more damage within 4 moves.

And checking this pawn break vs that pawn break etc. Human have thinking limit, depth 10, depth 20 etc. Some position require more detail analysis to compare between two moves ,beyond human calculation.

How about reading what I wrote under that section before excoriating me? Maybe you will understand that this post is regarding streams and not for personal use.

Honestly, if people just took the time to read what I wrote, they wouldn't be so offended.

Those armchair 1200's and trolls have made me partially an elitist (not really) in a sense.  It grinds my gears when people, who can't even avoid dropping pieces, can claim some position is so easy to play. 

Thank you for the good will as always Scott.

I don't think you even read the post. In my opinion, you just wanted to state your own opinion and be done with this post. Likely, I won't even take you seriously from now on.

Your first statement is an ad hominem and will be ignored. 

Your second statement is interesting and I find it strange that you state it like it is a fact. I would think it's the job of the commentators to let casual players know what is going on. I don't think the engine does a good job of doing that. The engine, as mentioned in the first argument, devalues what the commentators are talking about. You can read the example I gave there for my explanation. 

I do agree with the sentiment that it adds drama to the commentary though. Seeing an engine evaluation bar go up and down with the commentators' loud voices seems to make the stream look lively. However, when I think about streams with and without the engine bar, it hardly has an impact on the drama itself. It's just a bar that moves up and down.

As to the notion that it helps casual players follow the game, I don't think it really does. It tells what the objective evaluation of the position is, but not what the players are thinking, which is what really matters. That's the job of the commentators. 

While I do agree with you, this isn't relevant to the argument I laid forth above. Thanks though.

I remember you from long ago. You're not one of these idiots who shouts internet debating buzz words like ad hominem and strawman when they don't exist, so don't start now. My claim that it costs a lot of put on a production like STLCC is a fact even if I say it in a way that's a bit rude to Mr Rex Sinquefield.

You're right, I didn't read it entirely. For example just looking at the #2 headline ("it's distracting") and seeing the picture was enough to let me know that (at least that section) is nonsense.

So ok, I skimmed that section, and yes, confirmed, it's nonsense.

What I mean by the engine letting people know what's going on is, for example, I can tune in to nearly any sports contest and know what's going on even though I've never played it and I may not even know all the rules. Ball goes in hole / hoop / goal? That = good. That's all I need to know. Even a child could understand. Chess is necessarily (much) more esoteric, and the reason that world champions in chess earn less than 3rd rate sports players i.e. they're not as entertaining because chess is not nearly as watchable.

My point is that the engine eval can distract people from what's actually going on (just not because it's too much information).

Well, I for one disagree with the OP. I don't think it presents a problem the way Danny and Robert handled it. In fact it adds some significant tension and interest as they are going on about the position and the engine doesn't agree.

And anyone who knows a significant amount about computer chess engines also knows that their evaluations should be taken with a large grain of salt. They are not omniscient. They are interesting, at least to some of us.

At the end of the day, it is simply another independent take on the position, that may or may not be correct.

It is an ad hominem whether you like it or not. I call things for what they are. 

While I do understand the STLCC streams are costly and are funded by a generous donor, that is an irrelevant point to what I am getting at. The idea of discussing a position extensively and going over it with an engine later on is not crazy to think about for streams. You seem to be fixated on the thought that it would be expensive to produce streams in this format. Why do you think that Danny and Robert would not be able to follow that format? 

You didn't even read what I wrote again there. I gave a backstory that explains why our peripheral vision is instinctual and something we can not control. To claim its nonsense means you just chose not to read it, haven't thought about it, or aren't capable of thinking about it. Besides, you don't even explain why it's nonsense. 

Funnily enough, the last paragraph of your posts seem to have the most interesting talking points. The need for a scoreboard is an interesting concept I need to mull over. That's interesting. 

A much better criticism of my argument is that what STLCC does isn't relevant because showing the engine intermittently is not necessarily expensive. You seem to touch on this, good job, however what I argued is still not an ad hominem.

So ok, that's a good point. During the WCC I had a few tabs open. One was chess24, one was chessbomb (and a few others). I only checked in with the engine periodically. I mostly watched the engine-free chess24 stream, so I personally agree with you.

I just don't know if that's what e.g. chess.com is aiming for. If they want mass appeal I think it's reasonable to include the constant engine bar. While I can appreciate (or at least try) a GM position, a new player is going to be lost. I think for mass appeal it's nice to know who is ahead at a glance, just like a sports game.

As for distraction, first of all, a chess game doesn't move very fast, and neither does the analysis bar. If I were trying to watch a sports game and they kept flashing commercials on the bottom half of the screen, sure, I might become so annoyed or distracted that I'm not watching the game anymore, but this is a very awkward argument to make for chess games and their eval bars. It's like filler when your teacher told you to write a 5 page report and you only have 3, so you start talking nonsense to fill up more space.

Hi, thank you for your polite response. I appreciate it. 

Can you please explain why it "it adds some significant tension and interest as they are going on about the position and the engine doesn't agree" ? I'm interested in understanding how it adds tension and interest. Also can you explain how argument #4 doesn't apply here?

"And anyone who knows a significant amount about computer chess engines also knows that their evaluations should be taken with a large grain of salt. They are not omniscient. They are interesting, at least to some of us." 

I do agree with this statement because engines can only teach us so much. 

However,the engine's skill level is at a level where it is, for all intents and purposes, God to us by comparison. It is a higher arbiter and better decision maker than we can ever strive to be, in my opinion. 

"At the end of the day, it is simply another independent take on the position, that may or may not be correct." 

At least you sound open minded about this. Thanks for reading. 

What an eval bar may do, however, is like that 14 year old I mentioned. It may make them believe that all other analysis is superfluous. They're not distracted, they're disinterested.

Yes, many teens and younger kids have short-attention spans. We can not avoid dealing with their behavior. That's why I don't talk about this as there is nothing we can do about it, as you have said previously.

Your statement, "All you need is a billionaire octogenarian who has more money than he knows what to do with." is an abusive ad hominem as defined here.  At best, it's an insult towards Rex.

The need for a good scoreboard, as I mentioned before, is important. I need to think hard about this one. 

As for your last paragraph, the bar moves every time they make a move, which I would guess is every 7 to 8 seconds on average. So it moves frequently enough.  For quicker games and formats this is relevant. For longer games, games are slow enough where I don't think it helps to have the engine on display at all. The commentators can talk about a position for so long and tell us the evaluation after some thought effectively. 

The situation I described in the first picture, way at the top of post #1 was more problematic. This was about the Speed Chess Championships, where games are quick. The bar is moving every few seconds. It is disruptive and moves often enough where it is a problem. 

1.  OP thinks his opinion is so valuable that there is no limit to the number of words he can post that the audience will not find fascinating.

2.  I like to see the computer analysis, but agree there should be a way for people to turn it off, like they do on chessbase?

3.  F*ck Rex Sinquefield right wing radical prick who happens to like chess. 

An ad hominem is not an insult, a lot of people seem to misunderstand this.

An ad hominem is when you support your argument with a personal insult. Remember it's a logical fallacy. It's only fallacious because it's illogical, not because it's insulting.

For example if I said the eval bar is good or bad because Mr Sinquiefiled is a dummy, or a spendthrift, then that's an ad hominem. If I say the eval bar is good/bad because the cost is impractical unless you're a billionaire dummy / spendthrift, then that's not an ad hominem, it's just a rude way to make a point.

What's really funny is when the topic itself moves to a person's intelligence. Lets say the person you're arguing with claims they're really smart at some point, and there's some back and forth about that. In that case you can say "You're wrong because you're dumb" and it's not an ad hominem, because the topic under discussion is the person's intelligence. This really blows the mind of the keyboard warriors in the middle of their logic 101 class, or trying to sound smart after reading Wikipedia for 5 minutes.