Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GitHub Arctic Code Vault: Tech Tree

github.com

Post reply on HN
Print view View original

81–90 of 105 posts

Re: GitHub Arctic Code Vault: Tech Tree

#81

Post by threeseed »

fnord123 wrote: post #34
threeseed wrote: post #8

Some curious selections there: - 4 books on SQL but none on key/value, graph, document, time series, bigtable etc databases. - 3 books on C, 2 on JS but none on Haskell (leading FP language) or Scala (first to bridge FP/OOP). - None on search engines / algorithms even though it's the most widely used aspect of computing. - None on anything mobile even though it's the dominant computing platform.

>Haskell (leading FP language) Scala is the leading FP language (thanks to Spark, Kafka, Akka, and other massive projects). And TIOBE puts Scheme and Lisp ahead of Haskell. And Emacs Lisp probably has more installs than GHC.

Leading more in terms of influence than usage.

Most of Scala is derived from Haskell and even now the ecosystem heavily looks at what did/didn't work e.g. Cats, ZIO.

Re: GitHub Arctic Code Vault: Tech Tree

#82

Post by celticninja »

mumblemumble wrote: post #80
mm89 wrote: post #78

Earlier quoted context omitted.

I haven't read the document but what's wrong with the term "human"? Is there some other intelligent life-form on this planet contributing to technology and software today that I'm unaware of? Picking on the word human just seems nit-picky, like someone's not woke enough to use a more inclusive term.

My reading was that the complaint was that it implied false inclusiveness, which can come across as solipsistic to those who weren't included. A little bit like how the final stage of Major League Baseball's playoffs are called the World Series, even though 29 of the teams are based in the USA and the 30th plays in a stadium ~25km outside the USA. I imagine that is at least a little bit irritating to professional bas…

Except the World Series is named after the original sponsors of the competition and is not meant to infer worldwide participation. I'm not from the US and I am not from somewhere that plays competitive baseball and I know that,therefore I assume in the competitive baseball world this is known widely enough.

Re: GitHub Arctic Code Vault: Tech Tree

#83

Post by wheels »

pintxo wrote: post #66
jovial_cavalier wrote: post #63

Earlier quoted context omitted.

The core of the project is written in English, which is the current lingua franca of the world. If the purpose of the archive is for future scholars to have a corpus of knowledge from our time, then limiting it to one language maximizes the probability that the entirety of the information stored there will be recoverable. Including texts in Finnish and Hungarian almost ensures that parts of the archive will be lost 1…

If readability in 1000s of years of time is the goal, wouldn't you want to add as many (common) languages as possible? And for each book add as many translations of them as possible? That way they will be able to read most/all even if just one language survived - irrespective which one.

Languages don't maintain internal mutual comprehensibility over those sorts of time frames. This is what English looked like 1000 years ago:

https://www.poetryfoundation.org/poems/43521/beowulf-old-eng...

The only ones that maintain any significant population that can use them are those that are liturgical languages (e.g. Latin, Hebrew, Classical Greek, Classical Arabic, Sanscrit, ...)

Re: GitHub Arctic Code Vault: Tech Tree

#84

Post by hombre_fatal »

yorwba wrote: post #59
jstummbillig wrote: post #12

Earlier quoted context omitted.

I would be very interested to hear about important omissions due to that bias.

The "Fiction, culture, and history" section includes a bunch of works written in various languages, but apparently they're using the English translation. Surely the original is more representative of human cultures as a whole than its translation into English (which is more representative of anglophone cultures).

I think there's more to the story than just anglocentric oversight.

English translations are more likely to be unencumbered by copyright, and US copyright is often shorter (or more specific) than in other countries which can apply separately to translations. You can see this on https://babel.hathitrust.org/ where English translations are available but, say, Spanish editions are only available for search—you can't read them.

And because there are a lot more scholars using English, English translations tend to be more numerous and higher quality and more available than the original language. You see this a lot on Project Gutenberg where there's a super polished English translation and a non-existent or crappy original-language transcription.

For example, can you get a Spanish copy of "One Hundred Years Of Solitude" for this without legal issue? Maybe Harper Perennial was willing to cooperate with their English edition and others weren't? I don't think things are as obvious and straightforward as we like them to be.

I've run into all of this while "remastering" old Spanish works. English dominates culture, and not in a bad way. The only reason some new editions/transcriptions of old non-English works exist is because an English-speaking scholar was interested in it or a professor remastered an illegible edition to teach his Spanish language class. And now it's the only version of that work that's not behind a paywall. And you'll have to re-transcribe a messy scan of a book from the 17th century if you want a digital copy in the original language.

Anyways, has Github stated why it's English-only? Did they not have a good reason or are we just guessing that they didn't know that other languages are important (like we, HNers, people of culture, know)?

Re: GitHub Arctic Code Vault: Tech Tree

#85

Post by pintxo »

wheels wrote: post #83
pintxo wrote: post #66

Earlier quoted context omitted.

If readability in 1000s of years of time is the goal, wouldn't you want to add as many (common) languages as possible? And for each book add as many translations of them as possible? That way they will be able to read most/all even if just one language survived - irrespective which one.

Languages don't maintain internal mutual comprehensibility over those sorts of time frames. This is what English looked like 1000 years ago: https://www.poetryfoundation.org/poems/43521/beowulf-old-eng... The only ones that maintain any significant population that can use them are those that are liturgical languages (e.g. Latin, Hebrew, Classical Greek, Classical Arabic, Sanscrit, ...)

That would be an interesting task for some scholars: "We have a bunch of technology literature here, could you translate it for us into classical greek?"

Re: GitHub Arctic Code Vault: Tech Tree

#86

Post by fnord123 »

threeseed wrote: post #81
fnord123 wrote: post #34

Earlier quoted context omitted.

>Haskell (leading FP language) Scala is the leading FP language (thanks to Spark, Kafka, Akka, and other massive projects). And TIOBE puts Scheme and Lisp ahead of Haskell. And Emacs Lisp probably has more installs than GHC.

Leading more in terms of influence than usage. Most of Scala is derived from Haskell and even now the ecosystem heavily looks at what did/didn't work e.g. Cats, ZIO.

> Most of Scala is derived from Haskell

Scala had a lot of influences, but the first two papers on Scala have references to SML and Ocaml but none for Haskell[1]. Make of that what you will.

> Leading more in terms of influence than usage.

Is Haskell influential? By what metric? Impact factor of papers published?

Here are some features that Standard ML had which were cutting edge:

    strong static typing
    automatic type inference
    exception handling
    pattern matching
    parametric polymorphism
    first class functions
Basically all of these features are features any compiled language would enjoy today. SML is a very influential language despite not being used much in industry.

Here are features which Haskell is known for aside from the above:

    laziness
    strict immutability 
    do-notation
    typeclasses
    operator overloading with symbols like +++ and ==
Of these, typeclasses are wicked and influenced Rust's traits (the rest of the language being heavily influenced by [oca]ml). The rest are not even desirable. Haskell doesn't seem very influential in comparison. Maybe it's influential because it warns us to not default to laziness or go full strict immutability?

This is distinct from e.g. Erlang where the features like process management through the supervisor, mnesia, and so on are highly desirable for any developer even if other languages/platforms haven't implemented them as a standard component. (, vs . for line termination, not so much).

[1] https://www.scala-lang.org/docu/files/IC_TECH_REPORT_200433...., http://lampwww.epfl.ch/~odersky/papers/ScalableComponent.pdf

Re: GitHub Arctic Code Vault: Tech Tree

#87

Post by jovial_cavalier »

pintxo wrote: post #66
jovial_cavalier wrote: post #63

Earlier quoted context omitted.

The core of the project is written in English, which is the current lingua franca of the world. If the purpose of the archive is for future scholars to have a corpus of knowledge from our time, then limiting it to one language maximizes the probability that the entirety of the information stored there will be recoverable. Including texts in Finnish and Hungarian almost ensures that parts of the archive will be lost 1…

If readability in 1000s of years of time is the goal, wouldn't you want to add as many (common) languages as possible? And for each book add as many translations of them as possible? That way they will be able to read most/all even if just one language survived - irrespective which one.

There is a finite amount of space in which to store data.

Even if diversity of language is a principle that you adopt while creating this, not every single item in the database can be a Rosetta Stone-style snapshot of the state of human language in 2020.

Re: GitHub Arctic Code Vault: Tech Tree

#88

Post by jovial_cavalier »

ardy42 wrote: post #77
Bayart wrote: post #72

Earlier quoted context omitted.

Being trained as a historian, the idea of throwing away original texts in favour of translations because they're not in « the right language » is hurting my soul. If a text is lost because it's only written in Hungarian, it means the Hungarian language is lost. And that means not enough texts written in it were kept. See the problem ? Keeping as much linguistic data as possible is beneficial. Intentionally curtailing…

>> Including texts in Finnish and Hungarian almost ensures that parts of the archive will be lost 1000 years from now. Even if those languages are alive in 1000 years, the likelihood that interpreting period Finnish and Hungarian from 2020 will be possible is far smaller than the likelihood that interpreting period English from 2020 will be possible. > Being trained as a historian, the idea of throwing away original…

I thought the point of this was to preserve technical knowledge. It's not poetry. You can translate it from language to language and it should all be isomorphic, because there is something measurable and concrete underlying both expressions.

Re: GitHub Arctic Code Vault: Tech Tree

#89

Post by dgellow »

mumblemumble wrote: post #80
mm89 wrote: post #78

Earlier quoted context omitted.

I haven't read the document but what's wrong with the term "human"? Is there some other intelligent life-form on this planet contributing to technology and software today that I'm unaware of? Picking on the word human just seems nit-picky, like someone's not woke enough to use a more inclusive term.

My reading was that the complaint was that it implied false inclusiveness, which can come across as solipsistic to those who weren't included. A little bit like how the final stage of Major League Baseball's playoffs are called the World Series, even though 29 of the teams are based in the USA and the 30th plays in a stadium ~25km outside the USA. I imagine that is at least a little bit irritating to professional bas…

That’s a quite good analogy, yes.

Re: GitHub Arctic Code Vault: Tech Tree

#90

Post by oscargrouch »

jstummbillig wrote: post #12
dgellow wrote: post #5

The text uses the words "culture", "human history", and "the world/our world". But reading the list of sources included, it seems to mean "the English culture" of US + UK. That's a bit of a shame IMHO.

I would be very interested to hear about important omissions due to that bias.

No problem being mostly anglophone centric and english being used as a common ground, otherwise it would complicate too much everything.

But i have a problem with the lack of representation of great books and cultural achievements/standards that are not anglo-centric at all.

I miss a lot of great works of the human kind.

This is so important that they should have specialized people to curate that list and not just get "the list of great books that are on the top of your head when you only have an average capacity to do so".

Where is Cervantes, James Joyce, Kafka, Rimbaud, Pessoa, Homer, Goethe, Proust, Shelley, Voltaire, etc..?

It doesn't need to be that much inclusive of course, but it would be cool if it was a small window to the broader human soul instead of a subjective perspective that seems to be missing a lot of the common ground that helped to shape the civilization the way it is.

Post reply on HN