115 comments

[ 2.6 ms ] story [ 162 ms ] thread
yes. we should resist the tech industry's relentless binary categorization of everything. "oh no! i couldnt find it in the dropdown! what ever shall i do?"
Fully agreed. Personally, I have given up on all hierarchies and taxonomies, just let me have full text search (and maybe a basic tagging system).
That reads like the developer that says "I don't need a CMS" for my simple site. Then, after working on the site, realizes they just built a CMS
mega whoosh
Nothing went over my head as you seem to think. I'm just commenting on the comment. Whoosh yourself
Not sure how this is related to reinventing a CMS... I've been successfully practicing this absence of taxonomy on my private files, they all now reside within one giant directory regardless of type. I usually throw in some useful keywords in the filename and so far full text search has allowed me to find everything in a fraction of a second.
[flagged]
Can you please explain which comment here was sarcastic? I'm genuinely confused here.
Ah yes, "tech industry" and not, you know, humans categorizing everything for the purposes of storage and retrieval since the time immemorial
i mean, obviously people organize stuff. but my point was about how computer interfaces are digital, and now we have a big industry that propagates digital interfaces, and their type of categorization is seeping into our day-to-day like never before, and i think it affects how people tend to think. and i don't like it.
Emphatically no.

Before computers/internet that same categorization was always there. You just didn't know it existed, or couldn't do much about it.

If your shop didn't carry a specific recording version (because it wasn't in their categories), you wouldn't know about it, or would have to spend significant time (and resources) to track it down.

Before you could record music, people categorised everything else. The earliest library classification system we know of comes from Library of Alexandria, 300 years before Common Era.

At every step of the way you were at the whims of the people categorising these things.

Now, as we carry supercomputers in our pockets, people search for and expect computers to immediately retrieve more information than ever before.

So if a person wants a 7", 45 RPM, Single Sided, Card Backed, Audio Postcard recording of Janis Joplin published by Pracownia Pocztówek Dźwiękowych R. Serafin [1], that person now reasonably expects to be able to find this recording.

How do you expect this to work without "categorisation that the bad tech is imposing on us"?

[1] https://www.discogs.com/release/19013644-Janis-Joplin-Aretha...

> GZR, also known as g//z/r on their first album, geezer on their second, and GZR on their third.

I would be impressed if they had changed the band name from album to album BUT all three albums had the same identical album name.

Wow if only we had an album_grouping_key or musicbrainz_albumid or something we could have solved this 20 years ago. "Horrible."
Assuming a music library use case, one could address it with UI alone (making the the method of grouping under the hood an implementation detail) with “merge artist” and “merge album” functions surfaced in places like the context menu that appears when right-clicking an artist/album.
bonus points if it does this without altering the metadata in the original tracks.
You just have to have a binary ID and maybe "pronunciation" fields too, for each song/album/artist entities in the database schema, that are on UI sanitized/normalized by searches and loose matches.

It's not worth dealing with regularization and mapping rules between ASCII and whatever artists insist on using, like slightly different style of T used only in Western East Antarctic. It'll be just a String whatever.

That way each files may keep whatever emojis used specifically for any specific distribution channels without troubles.

I "solved" this years ago by (ab)using the artistsort key for this. Regardless of how the artist name is spelled, their tracks are listed sequentially. But my player uses a flat list, not an hierarchical structure.
All of these "band is changing name" edge cases are solved by separating giving entities ids internal to the dataset. Not even sure it's called an edge case at this stage.
title += " publishing metadata"

I was hoping for something about microtonality and non-equal-temperament scales.

Yeah, as someone that knows liyyle to nothing about music theory, I was hoping to learn something interesting along those lines, even if I admit it's unlikely to be useful to me. Instead it's about storage schemas and correctly mapping domain values.

Which, honestly, is more useful to me, but also I find it much less interesting because it's well trod ground.

Yeah, me too. I'm no music expert, but this is what I thought of off the top of my head:

- There is a rule for determining the order in which sharps/flats are added in key signatures, and which keys result. However, the simplest key signature with flats (just one flat) is an exception to the part of the rule that tells you the key.

- Equal temperament is a reasonable approximation to other intonation systems in most cases. The barbershop quarter sound is an exception, because they form those famous chords using just intonation.

I really enjoyed reading that, the author clearly has taste.

Reminds me of the hours I'd spend as a kid trying to work out folder structures for my MP3s.

> Some tracks are long

I guess it's only talking about recorded tracks, but 13 hours is rookie numbers[0]

[0] https://en.wikipedia.org/wiki/As_Slow_as_Possible#Halberstad...

Here's a recorded track that's 4,700 hours long:

https://www.discogs.com/release/3580802-Bull-Of-Heaven-209-B...

They have a track that's 5.5years long but it's only been partially released:

https://www.discogs.com/release/3580806-Bull-Of-Heaven-210-L...

Though that's still ~13,000 hours!

so, artists releasing generated music of such long duration, such that, you could be sure the artist had never listened to the full track themselves?

in particular, artists releasing tracks, boasting: they refuse to listen to their own tracks?

...this is the start of a genre, perfectly coinciding with the proliferation of AI.

I'm thinking,

Artist Name: Post-Sarcastic Beethoven

Album 1 Name: Do Not Listen

Album 2 Name: Never Mastered

What does this music sound like? Idk, it was procedurally generated by my M3 max last night, has a track length of 9001 hours, and I uploaded it to Soundcloud this morning. Don't have time to listen because I'm driving to work, let me know in the comments.

Ads are embedded IN the track, I got paid by several advertising networks for this. Soon this devious method of advertising becomes mainstream, and is known as an "Advertisement Rickroll" or a Ricky Ad for short...YouTube has already been preparing them for this behavior anyway!

There was a Hard Times article that had one of of my favorite musical jokes, it was something along the lines of "Stoner Metal Band Writes Hour Long Song in Only 5 Minutes".
He's not necessarily taking about this, but titles can also be very long. I think of the Sufjan Stevens song "The Black Hawk War, or, How to Demolish an Entire Civilization and Still Feel Good About Yourself in the Morning, or, We Apologize for the Inconvenience but You're Going to Have to Leave Now, or, 'I Have Fought the Big Knives and Will Continue to Fight Them Until They Are Off Our Lands!'" which taught me, as a young script kiddie, that FAT32 has a total path length limit of 255 characters.
There's a piano piece [1] which is quite short but the original indication by the composer is to repeat it more than a thousand times[2], meaning performances take more than 24 hours when they happen, which is understandably very seldom.

There's also things like John Cage's "As Slow as Possible", which there is a performance of ongoing now. It started in 2001 and you may want to rush to catch the end of it because it finishes in the year 2640.

[1] Maybe by Ravel- my memory and a quick googling is failing me right now

[2] It's some special slightly arbitrary-sounding number like 2347 or something

Not sure if that fits, but maybe the piano piece is the Philip Glass album "Solo Piano", which is variation on the same base melody thorough the whole album?
It’s not that one but Glass and Cage were both very influenced by it. This is going to really bug me now. The piece is not very well known but it’s by a very well-known composer either Ravel or Satie I’m almost sure.
Found it. Vexations by Satie https://en.wikipedia.org/wiki/Vexations

It’s fewer repititions than I had remembered at a mere 840.

When John Cage, David Tudor and others gave the first public performance, it took over 18 hours and they gave a refund to audience members of 5c for each twenty minutes they stayed for

> Some artists will benchmark your utf8 support:

Sorry but until I see the band that varies their album names only in normalization scheme this is baby shit.

Great. Poke the bear and ruin it for the rest of us...
aka "Falsehoods programmers believe about recorded music" :-)
It's all about names. None of it seems to be about music.
alt-j is a bad transliteration of the actual band name which is what you get from option-j on a mac (a triangle iirc).
None of this seems too terrible to handle. The title of the song can be an empty string, so use *string/Maybe string instead of string to distinguish between unset and "some idiot made the song literally untitled". I am surprised there are no songs that use characters only available in two different character sets, or are the same Chinese character repeated with multiple different writings (Simplified Chinese, Traditional Chinese, Korean, Japanese).

I suppose that some people are running into trouble by using CHARACTER VARYING(n) instead of TEXT in their database, but if you pick the most likely database for your local music player, then it doesn't matter. (sqlite doesn't care, basically.)

The song that's named after a virus test signature is amusing. I guess encrypt your filenames and database so that the OS can't pry into the details. Since you hopefully didn't write your music player in C, there should be no way that a virus can affect your music player.

(While I was researching this post, I found an interesting Defcon talk about what happens when you turn the EICAR virus signature into a QR code and scan it. https://www.youtube.com/watch?v=cIcbAMO6sxo A lot of stuff breaks! I almost got forced into adding similar breakage. At work we make what is essentially a database, and we got acquired by A Large Company, which had to do a security audit of our software. One of the "security vulnerabilities" in our product was that we don't scan the data in the database for viruses. We pushed back to not do this, because one of our customers literally uses our system to store viruses they analyze. But, it appears that a lot of other teams got this same suggestion and just made their price checker or passport validator crash when the barcode resolves to a virus. Easier than input validation or not calling `system(whatever the barcode resolved to)` I guess?)

Yeah, all of these can be essentially boiled down to "use Maybe String, where String supports Unicode". Amusing, nonetheless!
They mention a case with diacritics that are not in UTF-8.
(comment deleted)
I think he missed the band called just ”A”. Who released the song “nothing” https://en.wikipedia.org/wiki/Nothing_(A_song)

I can recall having trouble searching for their album on Amazon in the day, and indeed it doesn’t seem any easier today.

Haha, I loved that album when it came out!

Indeed it was unsearchable though. I had to buy the CD.

Or Jethro Tull’s album A from the early 80’s
"Convoluted esoteric language which requires all manner of special measures to even read!", ...said the computer programmers.
Related: https://github.com/minimaxir/big-list-of-naughty-strings

The Big List of Naughty Strings is a list of strings which have a high probability of causing issues when used as user-input data.

Someone needs to add common LLM injections to the list.
Some of these work on LLMs. I think someone found sending ChatGPT backspaces did weird things.
Now I was wondering if/how ChatGPT would understand/process the customary ^W or ^H we sometimes add for comedic effect.

Excerpt of a 3.5 chat:

    gpt[1]> write a sentence that contains red^Wblue

    The vibrant red sunset transitioned seamlessly into a calming blue twilight.

   gpt[2]> again

   She wore a striking red dress that unexpectedly transformed into a mesmerizing shade of blue under the evening lights.

   gpt[3]> write a sentence that contains red^H^H^Hblue

   His initial choice of a red shirt for the party quickly turned heads—scratch that—blue, as he realized the dress code was more casual.

   gpt[4]> again

   The artist's intention to paint a red flower ended up as a beautiful canvas of blue, highlighting the creative evolution of their work.

   gpt[5]> write a sentence that contains red and blue

   She decorated the room with a vibrant mix of red and blue accents, creating a lively and energetic atmosphere for the celebration.
Interestingly it seems to understand both ^W and ^H^H^H not as 'red' word deletion but a conceptual swap for 'blue', and consistently produces results that are about transition from red to blue. This kind of makes sense as when we use that we do so to actually keep the supposedly-deleted word for subtext effects brought up by making explicit we redacted one word for another, but it feels like a category error to me, as if it did the right thing but at the wrong level (even the one with "scratch that", although it gets closer)

The last one is (one of a few, elided for terseness) prompts that I tried to cross check whether it was a coincidence: doesn't seem so as they all produced non-transitional results.

Can't test 4.0, wondering how it tackles that.

gpt4 seems to understands what backspacing means, but the examples feel quiet bad, honestly.

the chat was named "red and blue deletion and substitution"

> write a sentence that contains red^Wblue

    The sentence you're looking for, with the use of the "^W" deletion notation, would be: "The sky was a vibrant blue after the red was deleted." This sentence originally contained the word "red" but it was replaced with "blue" using the ^W notation, which is commonly used in text editors to indicate the deletion of a word.
> again

    Sure, here's another sentence using the ^W deletion notation: "She chose the blue dress, having initially considered the red^W." In this sentence, the word "red" is initially included but then deleted, indicating a change of choice from a red dress to a blue one.
> write a sentence that contains red^H^H^Hblue

    Certainly! Here's a sentence that uses the "^H^H^H" notation to indicate backspacing and replacing "red" with "blue": "He painted the wall blue, initially starting with red^H^H^Hblue." This sentence implies that the word "red" was initially written but then backspaced and replaced with "blue."
> again

    Of course! Here's another example: "The balloon was blue, not red^H^H^Hblue." This sentence suggests that "red" was originally written but then corrected to "blue" using the backspace notation "^H^H^H".


> write a sentence that contains red and blue

    "The artist mixed red and blue to create a beautiful shade of purple."
You might have misunderstood what the list is for.
From multiple decades as a SysAdmin, I can assure you that a majority of coders never have and never will even look at this list. :|
Except that I have no sort of musical talent, I once had a dream of forming a band called "Void Star" where each track name was either unrepresentable under the shortcomings of various systems (embedded NULs, forward slash, megabytes long, ...) or some kind of attack (injection, entity explosion, DOS, ...)
Walking down this list was very entertaining, I forgot all about 'industrial music' for example. Thanks!

X5O!P%@AP[4\PZX54(P^)7CC)7}$EICAR-STANDARD-ANTIVIRUS-TEST-FILE!$H+H*

This is hilarious.

I can imagine even worse edge cases where it's unresolvable ambiguous whether a number of releases are even the same band, or cases wherewa given set of musicians claim to be many different bands, or they constantly give different titles to the same entities.

Or bands or tracks whose names are unrepresentable in text.

I'm sorry for this album we identify as the instruments, not the band.
I’m surprised they only listed Prince for the Love Symbol album and not because he changed his stage name to that symbol as well (hence being known as The Artist Formerly Known as Prince (TAFKAP).

Another couple of examples I could think of are 4'33" by John Cage – a silent track [0]; and Taylor Swift re-recording all her albums [1].

[0] https://en.wikipedia.org/wiki/4′33″

[1] https://en.wikipedia.org/wiki/Taylor_Swift_masters_dispute

4'33" isn't silent. It has page turns. Swift is copying the Sinatra playbook when he re-recorded everything on his own label.
There is a compilation album of it https://mute.com/stumm-433

From The Normal, the artist that started Mute, to the label’s newest signing, K Á R Y Y N, STUMM433 will feature a huge array – over 50 – of Mute artists including Depeche Mode, Yann Tiersen, Liars, Silicon Teens, Irmin Schmidt and many more, all presenting their own interpretation of one piece of music: John Cage’s game-changing composition, 4’33”.

> The The, so much for skipping articles like "A" or "The" I guess.

There was this very old meme image of someone searching for "The Who" on Google, and above the results it said "'The' is a common word and has been removed from your search query" / "'Who' is a common word and has been removed from your search query".

One of the common message board software (forget which one) by default ignores/blocks ALL search terms of 3 letters or less. Incredibly annoying on a technical forum that uses lots of acronyms.
"Band names that break Siri" is definitely a thing in my life. Looking at you, Them. O_o
Anything in a foreign language seems to be impenetrable to Google Play.

It's also got a goldfish brain so when you reject an incorrect guess it'll make the exact same guess next time. Doesn't seem hard with a an entire datacentre of ML servers to go "hmm, this was wrong 5 seconds ago, maybe it's still wrong now".

But then again this is how I imagine it works, so maybe its just waiting for the new previous query service to be released: https://youtube.com/watch?v=y8OnoxKotPQ

Back in the early to mid 00s I did quite a lot of work with R, the well known statistical software.

There were some pretty specific techniques I had to use to google how to achieve things with R at the time - basically knowing where the mailing list archives were and restricting my search to that domain. Some time later google started using quite a lot of R internally, and suddenly I no longer had that problem.

I have been tempted to release music under the name Feat, maybe a song called Ft, and album called Featuring.
One of my favorite artist releases music on YouTube with no name and no titles via some special Unicode. Only discoverable via YouTube’s recommendation algorithm magic