Hacker News Reader

Comment by simonw | Hacker News Reader

I was not at all impressed with the analysis in that thread. It makes a bunch of assumptions that don't feel very thorough to me, but announces them as if they are unimpeachable facts.

Biggest example is this one:

"9. Making up words or misspelling hurts - Words that are identified as “unknown language” are given 0.01, which is a huge penalty."

The code in the screenshot for that looks like this:

    // Boost (demotion) if the tweet language is not one of user's
    // understandable languages, nor interface language.
    optional double unknownLanguageBoost = 0.01

That doesn't match the description of "Making up words or misspelling hurts" at all!

View on HN

Yeah on the surface this looks more like "This user doesn't know German we should make German tweets much less likely to appear on their timeline" kind of thing. Just a complete misinterpretation, presented as fact without any supporting evidence.

I agree that seems presumptive. But of course twitter conveniently didn’t share the source code for how that factor is calculated. Turns out it’s hard to evaluate a big complex distributed system by looking at just one slice and no data.

interesting though, that comment directly says that if I tweet in english and my interface is in german then I'll have a serious problem. Which I wouldn't think is so rare. How many people have english as their second language and use it online? It feels like this deboost would hit a lot of people. Or actually, I have my interface in english and I often tweet in german. So that would hit me from both sides.

"One of the user's understandable language" presumably refers to the list of languages in the browser's Accept-Language header. If you are tweeting in English, you'd likely have English in that header.

It may or may not use the Accept-Language header as well, but this is a user setting accessible via the Twitter web UI-- you specify your primary interface language and can specify any number of additional languages that you understand.

There's also a place in Twitter's settings to show languages that Twitter has inferred that you know (e.g. following a German-language account and interacting with it almost certainly means that you know some German).

Ironically, sounds like a flaw of using Twitter as a medium. I'm sure he has more to say, or at least a more polite way to say it, but there's a character limit.

The systemic racism people are going to go wild about this

I agree with you (although I think one should be upset about this, and you seemed to imply that you disagree with the "systemic racism people").

This would directly penalize endangered languages and dialects, which is a tragic loss for linguistic diversity. I think a lot of people hoped the WWW would connect these small and dwindling communities, and even help revitalize them, but policies like this sound actively harmful.

If we're looking at it through the lens of systemic racism, it seems pretty straightforward as well – anyone who speaks a non-standard variant or dialect of their language will have less power to share their thoughts and ideas. They will also see less other people doing so, creating a feedback-loop discouraging the use of non-standard language.

> anyone who speaks a non-standard variant or dialect of their language will have less power to share their thoughts and ideas

If a recipient doesn't speak the language, the speaker doesn't have much power in the first place. This is noise reduction.

It’s systematic Monolingualism. Which even though this is something that has a big negative impact on me and my particular life, I am willing to first attribute to ignorance before discrimination. It’s just easier to make a product that follows a one country one language pattern. And the numbers of multilingual people are in the minority.

> I am willing to first attribute to ignorance before discrimination

I think ignorance is a fair answer for an individual like you and me.

However when you are a giant organisation, the organisation can be evil without anyone individually being so. In organisation ignorance and evilness is really hard to distinguish, and easy to conflate on purpose.

It is the job of the leadership to not be ignorant, or at least to hire someone who isn't.

Thats why AI ethocs teams where the correct move.

Most of the world is multilingual:

https://www.linguisticsociety.org/resource/multilingualism

Of course in the US this is not the case, and Twitter is a US based company that creates software used globally.

Headline can be: twitter shadowbans cultural appropriation - that should trigger everyone.

input youtube thumbnail of cat in the hat enraged "DR SUESS CANCELLED?! TWITTER WON'T COMMENT!" ragebait youtuber.

Reply on news.ycombinator.com