#metabrainz

they copied our dimple keys no problem

let me look, thanks.

alastairp: that step is already done!

kinda expensive, but I guess that's the nature of those keys

La Galaxia de las Llaves Carrer del Dos de Maig, 279, 08025 Barcelona

yeah, to be expected.

8% of MBIDs received are not the same as canonical MBIDs from our data set.

out of those 8% unmatched MBIDs, 40% are mismatched values, and 60% are NaN values. (nothing received from the server)

!!!!

amazing

mayhem: ^

mlhd MBID -> canonical mbid table

yep

only 8%??

and mlhd MBID -> recording name (text) -> mapper -> mbid

so, 92% of the time, only using the MB database (1 simple query) gives the same results as going through the whole slow mapper/detuning/search process

Beware tho, the sample size was only 10k rows!

I think it'd be really interesting to try and characterise these 8% (well, I guess ~4% since some are due to bad server respond) to see why the mapper and the canonical table give different results

sorry, got a meeting now - happy to continue chatting afterwards if you're still around Pratha-Fish

It's alright, I'll catch you later :)

alastairp: galaxia de Claus has some really friendly people, thanks for the tip!

mayhem: Which day are you expecting us to arrive in BCN?

Pratha-Fish: we can absolutely double-check your code to see what the behaviour is for mis-matched data

in fact, I've mostly been focusing on the results of your experiments so far, but maybe next week we can both take some time to look at your code as well and see if there is anything to improve

alastairp: Yes, I was about to ask you to do a code review anyway

For best practices, performance / readability improvements, etc

The code for the mapping is as follows:

I think it would be nice if everyone one (but reo) could be there for Friday morning.

alastairp: it turns out my keys can only be copied in one location and the folks galaxia claus were kind enough to explain that to me. (unlike the chap who couldn't be bothered to do his job at LM) So, I went over there and as soon as I told them which address the key was for, they were happy to copy it. for 23€ a piece. ouch.

lucifer: are you about to send your invoice? if so, I can wait and do all the ones from that bank at the same time...

mayhem: sorry for the delay, was afk earlier. sent now.

CatQuest: did you have a look at the newer version? Upon further feedback the defamation part has been taken out completely (as it's covered in the MetaBrainz CoC)

If you search for 'v2' above, and then there's a bit of further discussion

i havent see nany newer version thna the one you posted wich contan it

There was some more feedback from alastairp after but this is the version with Freso's feedback incorporated

alastairp: This file is a subset of the MBID mapping output. https://usercontent.irccloud-cdn.com/file/bKivy...

aerozol: Much better! And yeah, SEO manipulation is both covered by the CoC but also by the "promotional content" one in the new one, I’d say. :)

^ It's a set of all rows where a rec-MBID was returned by the mapper, but the MLHD_rec_mbid and the mapper_rec_mbid don't match.

Excellent - alastairp mentioned that he found 3. unclear, if anyone has any ideas

aerozol: If you want more nitpicky things… Instead of "asked" I’d probably say "expected". I’d remove the quotes around review. Point 2 I’d maybe make more general - like "no plagiarising or copying from elsewhere", since we also don’t want people copying from newspapers, magazines, etc. that are not websites. And I’d want to say "unless you are the copyright holder." - the "original author/owner" is not necessarily

the one who holds the copyright or has a license to propagate the work.

👌 I will drag my ass out of bed and make some changes

Maybe with a "if you’re not sure if you hold the copyright or are otherwise allowed to post it, don’t." added too.

I’m trying to think of examples of 3 that wouldn’t also be counter to 1, but I can’t come up with any.

So maybe between guideline 1 and CoC’s last point, that’s covered?

Which means that the guidelines is now only two points… so maybe it would be better presented as two paragraphs rather than a numbered list? :x

(Also just noticed a p too many in suppppport.)

I wonder if we still want to be specific on the things that come up the most - promo/SEO, even though it's already covered in the CoC

There's a strange tightrope with promo because alastairp confirmed that we don't necessarily want it all gone

My suggestion was "3. Promotional content that doesn't meet guideline 1. will be removed"

I feel like if we’re going with a "two paragraph" presentation we’d want to flesh it out a bit more anyway. I feel like I’ve added enough stuff for the copyright to make a proper paragraph (yay! copyright! :|), but for 1) it could include examples like "an artist reflecting on their work would be most welcome, but someone just copying promotional cookie cutter text to promote their music would not be."

Trying to find a middle ground...

I thought about lengthening it again too, but I think there's too many grey areas to describe all specific situations so it might be good to not describe any, if that makes sense. I'll have a play

"no primarly promotional content unles marked as such" for when we cando this...

bte aerozol you know you can *edit* irccloud pastes right?

😱

That's cool!! But I'd like to keep track of the versions tbh

sure, but for small changes like MusicBrainz -> CritiqueBrainz etc

natta!