Live data from Hacker News

Tell HN: Whole Yandex Git repository leaked

news.ycombinator.com

61–70 of 369 posts

Re: Tell HN: Whole Yandex Git repository leaked

#61
post #13

Interesting that most of the code comments are in English.

Lingua franca with all their contractors most likely.

Not at all. People preferred to use english to feel some connection to the world. That was mostly psychological. A Silicon Valley cargo cult if you wish.

Also the company always avoided any use of external contractors.

Re: Tell HN: Whole Yandex Git repository leaked

#62
post #59

% bzgrep "BEGIN PRIVATE KEY" \*.bz2 disk.tar.bz2:Binary file (standard input) matches drive.tar.bz2:Binary file (standard input) matches extsearch.tar.bz2:Binary file (standard input) matches ...

Nothing surprising. The development culture was shit back there.

Though I would expect these keys to be just some stub config values which allowed engineers to quickly run the shit locally.

Re: Tell HN: Whole Yandex Git repository leaked

#63
post #60

Earlier quoted context omitted.

Probably Yandex will start watching their infrastructure very closely, and go through all known attack vectors that haven't been prioritized before to fix them ASAP. Won't be a boon for OSS, any author would be idiotic to read stolen source code and then decide to create a OSS library/project based on what they learn from it.

> Won't be a boon for OSS, any author would be idiotic to read stolen source code and then decide to create a OSS library/project based on what they learn from it. This is naive. This generation seems very sensitive to the prospect of computer crime. The stolen source code will almost certainly be read, and if deemed novel enough will be turned into open source projects. It may be tough to figure out those projects a…

>This generation seems very sensitive to the prospect of computer crime.

You're right. I remember a time, maybe 20 years ago, when stuff like this would be generally appreciated by any community of hackers.

Re: Tell HN: Whole Yandex Git repository leaked

#64
post #63
post #60

Earlier quoted context omitted.

> Won't be a boon for OSS, any author would be idiotic to read stolen source code and then decide to create a OSS library/project based on what they learn from it. This is naive. This generation seems very sensitive to the prospect of computer crime. The stolen source code will almost certainly be read, and if deemed novel enough will be turned into open source projects. It may be tough to figure out those projects a…

>This generation seems very sensitive to the prospect of computer crime. You're right. I remember a time, maybe 20 years ago, when stuff like this would be generally appreciated by any community of hackers.

I think the problem is people are entering a matured industry, all they know is the professional buttoned up culture we have today. People have no further curiosity beyond working and making money in tech.

They do not have memories of a time when people tinkered around doing all kinds of crazy and possibly illegal stuff, just for fun, just to see if the could. Sad really.

Re: Tell HN: Whole Yandex Git repository leaked

#65

What are the short-term and long-term implications of this? I assume a drastically increased attack surface and potentially a boon for open-source development? Anything else?

In fact the damage would be mostly reputational.

I expect the code to be mostly worthless. There is just too much of it, it's poorly documented and, oftenly, just badly designed and badly written.

And the actual important data (index shards, voice models, all that crap) is not in these dumps.

Re: Tell HN: Whole Yandex Git repository leaked

#66
post #30

Earlier quoted context omitted.

They are still the most popular search engine in Russia, they have the best brand, etc. Intangible stuff like that is hard to copy. Running a search engine on the scale of Yandex is very expensive so I don't think that they are going to be replaced by some startup that copies their code and adds a few features. Probably a bunch of bugs will be found but that's not the end of the world. Apparently they handle images p…

Knowing how stuff like anti-spam algorithms and ranking algorithms work in order to abuse them is probably the much higher value here.

The code is lot less worthy than the data for the models which is lot better guarded and is not in this leak.

Re: Tell HN: Whole Yandex Git repository leaked

#67
post #42

Earlier quoted context omitted.

> Won't be a boon for OSS, any author would be idiotic to read stolen source code and then decide to create a OSS library/project based on what they learn from it. With the current geopolitical situation going on, is this really true? (From a western developer's perspective)

Even if Russia were at war with the world, it would still be unethical to loot its citizens' property.

It should be treated the same way they've treated royalties.

Re: Tell HN: Whole Yandex Git repository leaked

#68
post #13

Interesting that most of the code comments are in English.

So you can confirm they are legit? Anything look exciting?

While I personally haven't worked at Yandex I can certainly say they are legit. I have a lot of friends who worked there and a lot of intranet URLs in documentation are legit.

Re: Tell HN: Whole Yandex Git repository leaked

#69
post #51
post #19

Earlier quoted context omitted.

I mean they were pretty much forced to sell business to the government recently. Coincidence? Maybe…

what business and to what government? are you in your dreams?

i guess it's about recent exchange of their news service to vk. not literaly sold to the government

https://techcrunch.com/2022/09/12/yandex-news-zen-vk-sale-co...

Re: Tell HN: Whole Yandex Git repository leaked

#70
post #39

What are the short-term and long-term implications of this? I assume a drastically increased attack surface and potentially a boon for open-source development? Anything else?

Using stolen code in an open-source project seems like a bad idea.

Then we can train FancyCodeGeneratorModel on it to bypass this unfortunate licensing problem.
Post reply on HN