FA >>
(Whose error rate? Both?).
Hi Francesco,
Sorry - was your question serious?
I think naturally, the vast majority of the audience are interested in gathering insights into *their own* error rates, rather than a robot's.
I'm sure we are all aware that robots play robotically and very, very well: in my experience consistently close to zero ER for 'standard' games with very low-variance.
Most of the value I/others get from match analysis is from going back over my/their own inconsistent (human) games/errors.
So, only the human's error rate please for MVP. Also has the advantage of keeping the filename as concise as possible whilst maximising the most valuable information..
Thanks also to Isaac Keslassy and Ian Shaw for supporting the idea. Hopefully it can be implemented soon.
------------------------------
Message: 3
Date: Thu, 18 Jan 2024 13:10:55 +0100
From: Francesco Ariis <fa-ml@ariis.it>
To: bug-gnubg@gnu.org
Subject: Re: Improved gnubg filenaming convention - suggestion
Message-ID: <ZakVT4LKM2oOemlE@x270.casa>
Content-Type: text/plain; charset=utf-8
Hello Joseph,
Il 18 gennaio 2024 alle 11:33 Wayne Joseph ha scritto:
> To propose a relatively 'easy' solution to this, I think it would make
> better sense to add an hour-minute timestamp to the existing file name
> convention?
>
> [snip]
>
> In addition, for a cherry on top, it would be even better if the error rate
> for each analysed match was also appended to the file name i.e. -4.3 or
> -8.5 or whatever. This could allow for even easier searching and
> filtering/foldering of match files with blunders for retrospective study on
> each user's PC
I find the suggestion to add full timestamp reasonable, error rate less
convincing (Whose error rate? Both?).
Tangent: GNUbg has some facility to store statistics in a relational
database, maybe it could help in your case.
—F