How is the new analysis system? of chess.com

flashlight002

Hi guys. Chess.com has communicated that there are accuracy problems with the new analysis engine. If you read my posts and ignobles posts on this post topic and in other topics on the forums you will find what I found out regarding accuracy issues. They are busy working on all this. It is possible that they may already have made some corrections and we will start to see improvements already. 

flashlight002

@LionVanHalen none of us will EVER be as good as a chess engine on full strength. EVER. Running Stockfish 10 on another app (not chess.com) I was seeing it operating at about 350 Kn/s......on my phone!! That's 350 000 half moves per second!! Yes it is quite possible that an engine will calculate a move that makes no sense to one at that moment in time....because it is part of a broader strategy of moves it has calculated say 20 moves deep. However, many times the engine has shown me quite logically why I made a blunder and the better move to make. Engines are tools for us to use, but yes we need to use our own brains too to figure things out too happy.png. An engine is an aid. There are many times I scratch my head at why an engine made the move it did...but thats because I am not operating at 350 000 nodes per second and seeing 20 moves deep happy.png . But there are times when I see perfectly the logic of the engine's move. And then I have learnt something very valuable!! So don't discount the value of engine analysis. Much of the time it is very enlightening.

However one cannot deny that chess engines have contributed much to the body of knowledge on chess and how to play the game.

Kallatroh
LionVanHalen wrote:

Well... how come other post are saying the analysis is wrong and needing fixed?

Some of these opinion are plants?

Guys who work for the site, leaving positive comment, propaganda...

If you are talking about me, I am just a random player. I really think the previous version was almost unusable. 

flashlight002

@Kallatroh I think the new version better addresses how to learn from an engine's analysis. And that is key. I agree the new version and logic of the GUI is much better. I especially enjoy the retry section and the themes section is also very clever as it helps make sense of strategic and tactical motifs one may not have been aware of that contributed to certain positions or move decisions. 

The histograms are also a good learning tool as one can hone in on problem areas easier and understand the bigger picture.

All they have to do is fix the accuracy issues..as some of the errors the engine was making were severe! (See my screenshot of one example much higher up in this post topic). 

flashlight002

@ LionVanHalen I agree with you that one would have thought these accuracy issues would have been caught in the beta stage...but not much  we can do about that now. Things like this happen all the time in computer programming. One tests and tests and misses something in plain sight....or one tests on one set of hardware and then makes it live on another set of hardware and problems happen...or one opens it up to more users...causing a greater load on computing resources and problems suddenly happen.

The main thing is chess.com has acknowledged the issue and are fixing it. That's the main thing. Plus they want the initial scan that everyone gets...paying and non paying members to be accurate. Paying members will have more configuration settings but that one has to understand. Paying members expect something for the money they are paying. 

jurassicmark

Which is why I'm no longer a paying member after many years.  I wasn't getting what I expected.  I worked with a guy who said, "change is always good."  It's one of the dumbest things I've ever heard anybody say.

A_Flock_Of_Sheep3

I like it a lot.

flashlight002

@jurassicmark I could not agree with you more. Change for the sake of change is a flawed way of thinking and functioning. Sure the world is always changing in multiple ways, but change in my mind should always be for the better. If it is not then it is a very stupid and misguided decision indeed. In the business world at the strategic level I have seen this often. Competitors rush out new features or benefits in the name of change without either fully testing the changes (and thereby damaging their brand equity) or misreading their target markets completely and creating the wrong types of change (the outdated we know what you need vs. finding out what customers need and want). Often "change" may also be used as a form of "investor stroking"...look at how cutting edge our brand is...look at how we are constantly evolving, but the change is either "bad change" in that it reduces features, benefits and utility, or disenfranchises parts of the consumer base. 

But "good" change…that which has been properly thought out and properly implemented...and this is the most important point....change that better meets customer wants needs and expectations...well that can only translate into a raising of brand equity.

But one should NEVER just change for the sake of change. 

notmtwain
flashlight002 wrote:

@ LionVanHalen I agree with you that one would have thought these accuracy issues would have been caught in the beta stage...but not much  we can do about that now. Things like this happen all the time in computer programming. One tests and tests and misses something in plain sight....or one tests on one set of hardware and then makes it live on another set of hardware and problems happen...or one opens it up to more users...causing a greater load on computing resources and problems suddenly happen.

The main thing is chess.com has acknowledged the issue and are fixing it. That's the main thing. Plus they want the initial scan that everyone gets...paying and non paying members to be accurate. Paying members will have more configuration settings but that one has to understand. Paying members expect something for the money they are paying. 

The accuracy issues were reported in the Beta stage and they said they were looking into them.