Fixing New Analysis

@erik....my lips are sealed . But looks like the forum regulars have sniffed out my post. Oh well. At least they have a pre release teaser to get excited about now I guess! Lol. 

On a serious note... please please  give some serious thought and consideration to what I have written in the prior posts where I have  tried inordinately hard to convince you of the merits of my arguments, and that I can somehow change your mind and decision re the icons. More I cannot ask for I guess.

I look forward to your insights. Should you change your mind re the icons I am sure there will be many who welcome it back, and applaud you for your thought leadership and being close to your customers.

@erik after looking at a number of game analyses and the "Best Move Diff" graph that now shows colored bars...I have picked up that even the base line is picking up the move quality colour of a move that matched the engine best move. The colour on the thin base line changes. Here is an example of what I am talking about.... because I have A GOOD IDEA...so hear me out:

As you can see on the base line there was a change to blue to show an excellent move, and a change to green to show a "best move.

If you are steadfast on not bringing back the icons to the score graph then can I propose an idea to show up move quality colours for 0.00 base line scores better, because right now they are VERY DIFFICULT TO SEE as it is a thin line of colour : make the base line thicker so that it is then easier to see the move quality colour for a move that falls on the base line. Moves that fall into the brilliant, best and good categories are likely to always sit on the baseline, as a thin thin line of colour at present...which is very difficult to see. I have shown my idea above of thickening the base line.

For the record I still believe putting the icons back on the score graph is the way to go, as it was originally or with the clever idea proposed of white's icons just above the base line, and black's below. Then there would not be ANY problems like in the picture above that we now have.

https://www.chess.com/forum/view/community/not-sure-i-should-trust-the-basic-analyzer?

Mate in 3 seen by computer is called a blunder.

@notmtwain thanks for posting this. Wow that's a real shocker. @erik with outputs like this it really erodes trust in the new system unfortunately. Can your team explain why such a patent analysis error is being returned?

@PawnstormPossie that indicator showing depth = 10 is a real mystery. The poster shows an analysis mistake returned using the engine feedback system, which can only return an answer after reaching d=20. The move quality indicators ?? are shown on the board....and they only come up at d=20. I tested this many times. 

I can't reproduce this faulty analysis. See analysis of game, https://www.chess.com/a/2vESPex2SgQPk and click  on Ng6+ and it instantly indicates best move.

@jas0501 not the first time this has happened where a user shows the bizarre engine feedback result, but when everyone goes and tries to reproduce it they cant. Notice that your d= indicator says 20 and his in his screenshot of the bug showed d=10! My educated guess therefore is that the engine gave it's result at d= 10 when it should have only given it at d=20. In the original op thread @jdcannon has handed on to dev guy to investigate. But the above is my take on it. An analysis at d=10 is bound to give such a solution. 

@dallin @erik the bug shown and discussed in posts #358 #359 and #360 is still not fixed. @dallin can it be fixed please...or is there a problem preventing resolution? You promised it would be fixed "shortly". That was 8 days ago guys. Hence I am following up. Must I log a ticket with membersupport@chess.com? Please let me know. I don't want to come across as "nagging" or being a pain in the ..... but I would just like an indication of when it will be fixed.

@erik this just got posted here https://bit.ly/2GwfAjb

about the new analysis system. Not what one wants to hear from one's market! Far be it for me to tell you how to run your business, allow me to, as a loyal customer, and passionate marketer, give my take on this. The way I see it, the best way to change such deleterious market perception is to introduce various levels of engine analysis (so there is a basic, a mid and a high depth option…thus offering various levels of "accuracy"), and to improve current accuracy and output stability and reliability to the system as a whole, coupled with pr around all this work. A poll (with hopefully enough replies to make it a statistical relevant sample) may be an option too to see what your market perception is of the analysis system vs all this anecdotal evidence popping up all over the place. Just my thoughts.

That's exactly my point...  the analysis page should be the same in form and function regardless of the means by which you navigate to it.  An analogy:

To get to the closest hardware store, I can take the freeway or I can take surface streets.  Regardless of how I get there, the store will have the same number of entrances, and they will all be in the same location.  But suppose the number & location of the entrances changes depending on how I drive to get there...  This is how the analysis pages currently differ from one another - depending on how you navigate to them is as if number and location of the hardware store entrances is different depending on how I drive to get there.  Take the freeway = 2 entrances, 1 one the north, 1 on the west.  Take the surface streets = 1 entrance on the west.

Further, I think the functionality of being able to create a unique URL is brilliant.  But users can't do that if they navigate to the analysis page via the "analyze" button on a completed game page...  You have to copy the pgn, navigate to a blank analysis page, plug in the pgn, and only THEN can you create the unique URL.  It makes no sense to force users to go through all those steps on one version of the analysis page, when clearly the functionality is available to do it more quickly and more easily on another version of the analysis page.  Continuing my analogy... 

I like to use the northern entrance to the hardware store because that's where the 2X4s and plywood are located and those are the things I buy the most frequently.  But suppose I can only use that northern entrance if I drive to the store via the freeway?  Well, that would be unnecessary and inconvenient if I want/need to use the surface streets.  I'd either have to a) use an entrance that is inconveniently located to the goods I want to purchase, or b) ALWAYS drive to the store via the freeway so that somehow, magically, those northern are there for me to use.  It makes no sense.

Yes, I understand that the game report is run automatically when you click the "analyze" button from the completed game page.  But that doesn't preclude the possibility of a) greying out the "report" button (my preferred northern door has already been opened for me by a friendly employee) and b) giving users the same "unique URL generator" capability that's available from a blank analysis page (that western door is always there for me when I want to just dash in for a gallon of primer). 

One page with the same form and functionality, regardless of how users get there.

- M

(p.s. - But even more important is ensuring that the analysis is accurate...  So given the choice, I'll take accurate analysis over ease of use...  eg, quality lumber over preferred entrance.)

@mgt3 wrote:
Further, I think the functionality of being able to create a unique URL is brilliant.  But users can't do that if they navigate to the analysis page via the "analyze" button on a completed game page...  You have to copy the pgn, navigate to a blank analysis page, plug in the pgn, and only THEN can you create the unique URL. 

This is not the case. Just doing an edit and then a share and the address provided is a short url.

I've reported the bug that the need for an edit should not be a requirement, The address in the share should always be a short url.

Thank you for 1) pointing this method out to me, 2) submitting the bug report and 3) reinforcing my point - unless users specifically want/need that pop-up dialogue (eg, create a puzzle, embed, etc), it just creates unnecessary steps to get to the same result that can be more easily achieved by making the pages consistent.

- M

There is a subtlety that exists regarding the browser address. The one rational for having the browser address be analysis/game/whatever instead of the short url is that it is the analysis version of the file. Another user following that address invokes the analysis rather than viewing the owner's version. The users' edits are not visible via the analysis/game/whatever address. Private annotations vs. public annotations via the short url. 

I can show someone a game and the analysis report without revealing any annotations using the analysis/game address or anyone accessing via an archive analysis view. The act of doing the share does need the short url, that access revealing the annotations.

So I would lean toward having the browser address remain  analysis/game/whatever and as the bug report indicates always have the share address be the short url. 

A non-issue:  If you want to share your private analysis, share the short URL.  If not a) give the "analysis/game/whatever" link or b) just give the link to the game itself.

- M

@mgt3, my point exactly. Thus the browser address reflects the type of access and should be different, short url for a shared access and analysis/game/whatever for a non-shared access.

lol

1.) "Missed win" wouldn't make much sense, even if it was correct

2.) It's obviously incorrect (cf. -66, vs. M21 vs M7)

Pretty much the same bugs I posted months ago. (Nonsensical categorization and server/client contradictions)

@fschmitz422 I wonder what the dev guys have been doing to fix things. I am sure they have been fixing things...but why then the same types of issues cropping up I ask myself. Are they struggling to fix certain issues/bugs/scenarios? Have they failed to isolate the problems correctly? Can't they fix it in its current configuration?? I can't answer these questions...only chess.com dev can. And I hope someone from chess.com steps up to the mike and responds on this. I agree with you these kinds of results shouldn't be happening so many months after we started documenting these situations. I have seen coms where chess.com Support say "dev guys looking into this"...but we never get feedback that the problem was xyz and we have done xyz to fix it. So it's a bit one way. We report...and we get acknowledged (sometimes) but we don't get told its fixed.

So the ball is in chess.com's court to engage with us on why we are still seeing problems with this engine analysis configuration so many months later and what will be done to fix it. Plus there was also talk about enhanced analysis features (greater depth analysis etc) but nothing has transpired yet.

https://www.chess.com/analysis/game/live/3949529887

Thanks everyone. We recognize we have a way to go here still. We have some updates in the works - they are just not fast.