A bug that hasn't been fixed in 3 months.

Hello! About 3 or 4 months ago, I posted a forum post: https://www.chess.com/forum/view/site-feedback/a-possible-bug-in-insights

Basically, that post mentioned a bug in insights, where I look in insights under "Accuracy in Game Phase" and I only played one game recently, so I know what color I played, but the insights says I'm playing the other color, which makes me very concerned about whether I can trust the insights. 

So I took another break, I came back again. And I used the free trial to access insights again. I played a game as white. I set the insights to games that I've only played in the last 7 days because there should be only one game I played, which was me playing white. Insights under "Accuracy by Game Phase" says I played as black! I'm worried about insights now, because I feel like I can't trust it. 

I won't be paying for a diamond membership until this gets fixed, because I'm not paying for something that's broken. (Insights was the only thing I used on a Diamond Membership)

I actually looked at your Insights and it said the game you played yesterday was played as white.

I didn't see any games listed under the Accuracy by Move Number. It didn't look like you had run the accuracy report. I don't know if all the fields are updated immediately or run on a batch basis.

At any rate, I see that Martin Stahl forwarded your original forum post report to the staff. In the future, you might just append a post to an original thread to try to get something updated.

You didn't scroll down enough. You're looking at "Games that ended in..." I'm talking about "Accuracy by Game Phase"

I suspect your "Bug" is related to only one game, played as White, is in your database. You have no games played as Black so it basically is undefined (NaN - in computing means Not a Number).

Perhaps your bug would disappear if you played more games and had a least one game played as white and at least one game played as black in your database of games.

Um maybe? You realize I'm not complaining about the NaN right? My issue is with the part with the Accuracy by Game Phase right? Where it says I played a game with black, even though I was playing white. But I guess I can try playing another game if you think it would change that.

Yes I understand.

In computing, undefined variables fall into the category of Not a Number and they can cause a cascading effect of errors in a program. The fact that NaN is being displayed is "proof" there is a bug in the Insights programming. Experienced programmers know not to use undeclared/undefined variables in a program. The programmers did not consider the case of a one game database. It should be a fairly easy fix for them to fix this bug, but I imagine bug priorities dictate what the programmers work on.

Alright, i played a couple more games, and I think the issue is still there.

I played 2 games of white, and 1 game with black.

First Photo is me playing 2 games with white, the 2nd is playing 2 games with white, and one game with black. I think it did solve the error in "Games that ended in the..." part. But the issue is still there in "accuracy by game phase". It still think the 2 games I played as white are black.

It's possible your Insights database was corrupted by the NaN undefined variables from when you ran Insights with only one game. It's also possible that it may correct itself in a day or two, but I'm not sure.
If your Insights database is permanently corrupted, someone at Chesscom should be able to fix it for you. You would need to submit a Bug Report Ticket and be sure to tell them about our discussions in this forum post.

If you mean the lack of endgame accuracy, your games may not have reached the endgame phase in order to have an accuracy, based on the site's definition of endgame.

No, that's not my issue at all. I've mentioned it multiple times. The insights under "accuracy by game phase" think I played two games as black, when I played 2 games as white. It's switching colors. That's all.

That's not at all what's being shown. The Accuracy by game phase does not include a game count

The graph breaks it out into white and black games

I'm not assuming you're only looking at a screen shot. You really need to look at my comments before replying. This is my last 3 games. I've played 2 games with white, and one with black. Yes, they don't have actual numbers. But if you look at the accuracies, it's taking the accuracies of the 2 games with white and calling it black. Look at the 2nd screen shot i posted and compare it to the 3rd. The 2nd shot is when I only played 2 games with white, and they're showing up as black. The 3rd screenshot I played 2 games with white, and 1 game with black. The accuracies for black don't change, because it still thinks the 2 games I played with white are black. The screen shot are based on the games I've played in the last 7 days, which at the time of writing this comment, is only 3. I played TWO games as white, and ONE game as black. The insights under "Accuracy by Game Phase" think I've played ONE game as white, and TWO games as black, which is messing up my insights.

Or you can just look at my first screen shot. In this screenshot I had only played ONE game (as white) in the last 7 days. In the insights, it claims I played black.

Here's my view.

On Nov 19 one game was played as White, Insights thinks it's Black

On Nov 20, another game was played as White, Insights thinks both games are Black.

Also on Nov 20, another game was played, this time as Black.

Notice the Accuracy by game phase percent for Black has not changed from the picture directly above this one. Therefore Insights must still think there are two games that were played as Black but clearly were not.

What I don't understand is if Insights thinks the first two Games were played as black, what game is it showing for White since the last game was played as Black and added to Insights last?

Thank you for explaining the situation!

Ok, I restricted the view down to a single day and it's clearer now. I'll find out if it's a known issue.

Thanks Martin.

I think the problem is the result of Insights was not programed to handle a single game, (thus the NaN error being displayed), and Insights assumes there would always be at least one Game played as Black, and one Game Played as White to analyze.

Since I mostly play one Game a day and I wanted to document this bug using my own data, I've come to the conclusion that the piece colors for Black and White are being displayed reversed when viewing Insights Accuracy by game phase - by Piece Color..

On Nov 13, I played a game as White. Insights shows this game as Black (see the red circled data).

On Nov 14, I played a Game as Black and Insights added this Game as White when I view Insights for Nov 13 and Nov 14, (see the green circled data).

On Nov 15, I played another Game as Black and Insights added this Game to the White Game Data when I view Insights for Nov 13-14-15, (see the green circled data percent values increase).

Thank you Martin for the tip on viewing Insights using Custom Dates!