(Oh, and there are a lot of these people on HN who think writing garbage blogspam for their latest project is a great "growth hack". A hearty fuck you to you, too.)
(Oh, and there are a lot of these people on HN who think writing garbage blogspam for their latest project is a great "growth hack". A hearty fuck you to you, too.)
This happened to Google already with translation. Their translation system started with documents with known translation (e.g. EU and UN official documents). Then they started to use documents on the web that offered their own translations...until Google translate got good enough that people started using it to translate their own websites.
This "inbreeding" or bad feedback made the translations worse until this was figured out and I believe they long gave up automatically using translations found in the wild to improve training. Source: friend who worked on GT at the beginning.
It's harder to filter out competitors bad translations though...
This summer I had a lot of trouble finding information for something specific in Angular on how to do some specific thing, where the official docs didn't cover it in depth enough, and then searching for it online was impossible as it was thousands of "blogs" with the same copy pasted tutorial with a few key words changed around that show the same hello world example.
We often talk about size of the community as an important consideration when adopting programming tools. But really there is a sweet spot, where a community has enough serious people to produce the right information but not enough unserious people (yet) to bury it. Communities at this stage can be better than those for universal standards.
For Java I now rely mostly on some more reliable source for information that I tend to directly search rather than go through Google/DDG. Baeldung and Mkyong come to mind.
Surprisingly, some discord groups have also become my go to source for help.
Popular things like Java or JS have page after page of pure crap. Rust or Haskell will have generally pretty good content. Python used to have decent content, but has been trending downwards sharply.
A lot of it is due to excited beginners wanting to write about $topic. I don't think this is necessarily a bad thing as such; and I'm a bet less cynical about it than some others in this thread. It's pretty neat that anyone can write something and put it online. I just wish there was a way to filter this kind of stuff.
This is where the "throw everything in a big bag"-model of the web comes with some pretty big downsides. But I don't really know how to do it better either.
Gave up the inherited CI and rewrote everything in playwright in few weeks.
Many times I look up something, and all the YouTube videos and tutorials are telling the exact same thing and it's clear there was no effort at all or even if the author knows what they're talking about or even tested that code at all.
Many people outside of the industry don't know it, but so-called "copywriters" are people who are paid around 1-1.5x minimum wage with the expectation of being able to write texts on ANY topic. I know a girl who does this. Almost all of texts are bs and she is a top performer in her company. She does it because it makes bit more than being a waitress with bit less effort (and given covid, she'd lose a job if she was a waitress, and "copywriter" job is quite secure).
Shall we get surprised with the results?
By the way, a question - bullshit answers on Quora from paid posters - who's paying them and why? Or you mean those Putin trolls and wumaos? Their content tends to be very biased, but of decent quality...