I watch about 50% Japanese content and having to switch this off manually has become a major source of annoyance. Bizarrely, if I misunderstand something in Japanese and want to go back to check, these same videos normally don't have any English subtitles available. There's auto-transcribed Japanese subtitles which are about 90% accurate, but they're rarely translated.
The absolutely wild disparity in compute required to translate the Japanese text to English vs rendering an entirely new soundtrack in English blows my mind. I guess someone at Google thought it made sense because many people prefer dubs to subs, but that's on highly polished entertainment product vs 1-person-and-their-Japanese-vlog channels which are not aiming at a mass audience.
Great analysis. It's especially ironic to see it play out in this context given the well-known Japanese predilection for building consensus and buy-in; the flip side of this is that those outside the circle of decision-makers are especially sensitive to the subtexts you identify.
This style of communication certainly has its uses, and I too resort to it when I want to indicate firm disagreement without being aggressive, as do most people. I think the reasons it has generated so much pushback on this occasion are twofold: it's being used to dismiss the concerns of whole community by infantilizing a long-time leader of said community, and it's doing so in the context of translation itself. That is, volunteer effort and tools that are supposed to improve communication and mutual understanding in theory are in practice being replaced unilaterally with a tool that epitomizes a unilateral and dehumanized approach to information processing.
Right. IF they were taking the issue seriously the response would be more along the lines of 'we didn't intend this outcome, and have paused availability of the bot while we re-examine the issue.' Mistakes do happen in large organizations where communication and project management is highly distributed, but but appears to be less of a mistake and more of a fait accompli, where a decision was made at a high level to roll out a new feature and objections are being treated as a PR problem.
Data raw material is meh, as are the results (clubs are clustered by geography and music style? Shocker), but the post isn't about collection or survey methodology.
The hard work here is in the analysis and presentation, which are nice. That will still be valuable if applied to better-curated datasets. I don't blame the author for working with a mediocre dataset that happened to reflect his interests, he's not making big claims about what this means for the entertainment industry or the future of culture.
The Libertarian party got ~4.5 million votes in 2016. Getting some of those votes, or dissuading them from voting, is enough to make a difference in a tight race. See my other answer upthread for more context.
Both are true. For a long time the Libertarian party was seen as drawing away small numbers of protest votes from the GOP, being populated by (mostly) guys who rejected Democratic over-regulation and nanny-statism but also rejected the GOP's anti-abortion politics, criminalization of drugs etc.
A few years ago there was an organized effort to capture key roles in the Libertarian party and focus the organization more on property rights and capitalism, with less emphasis on personal freedoms and constitutional limitations on government. This effectively split the Libertarian party, neutering it as electoral factor.
Now, the Libertarian party never mustered a large share of the vote, but many electoral contests are won at the margins. They managed to get ~3% of the vote in 2016, but lost >80% of that over the following 2 elections.
Would love to hear more about this if you're inclined to share or have written about it somewhere. Legal contacts betweent he fedreal government and individuals are often surreal.
I read this, and found it to be a disappointing read. It had few details, and instead was more of a social sciences paper, covering basic ideas in academic language.
Roughly it seemed to be suggesting that:
* It's easier to deceive someone if they first solicit for help on a forum
* You can trick someone into revealing sensitive info like which infrastructure provider is used by nerdsniping them: "My mate thinks you should just enable health checking on AWS ELB", and then they reply "Well actually I use Hetzner". Except I'm guessing it was more elaborate than that.
I guess I wasn't the target audience of the article though.
joshmn, what did you think of the article?
Do you find it difficult to trust random commenters online now?
I see you mentioned you can't discuss technical details, but if/whenever that expires (?), that'd be great to hear.
I also found it underwhelming, though I'd like to think I’m the most scrutinizing of the subject matter. There's some nuance between my take on my behavior and the profiler's, but I'll give them the benefit of the doubt—they only had my Reddit posts to go on and had to package that for investigators.
I still tend to trust by default and make witty comments or jabs that sometimes land flat, so the article was accurate in that sense.
As for talking to the undercover, I made a point of keeping no secrets about my site's technical implementation. Between me and some "competitors," I was usually the first to respond to upstream provider changes—I'd even share my findings without expecting anything in return. Anyone could’ve asked about my issues, and I would've told them.
Trust is the most valued currency in the piracy world, and I worked hard to earn it with both peers and customers. Acting otherwise would've gone against that—and against my own morals. My being neurodivergent may also be worth noting in my willingness (or unwillingness from a free-will perspective) to trust others.
Technically speaking, the site worked by reverse-engineering the league's official streaming services—a few curl requests, careful observation of responses, and adapting them to my needs. There's more to it, of course, but my 2016 MVP was barely 50 lines of Ruby and a plain HTML file. TorrentFreak got some of the details right.
You keep arguing that not displaying a falg is justification for killing people. This is bullshit. It's justification for interdicting a vessel but not for simply blowing it up.
The absolutely wild disparity in compute required to translate the Japanese text to English vs rendering an entirely new soundtrack in English blows my mind. I guess someone at Google thought it made sense because many people prefer dubs to subs, but that's on highly polished entertainment product vs 1-person-and-their-Japanese-vlog channels which are not aiming at a mass audience.
reply