News for hackers. Stuff that matters.
Re:Tell HN: Whole Yandex Git repository leaked by imglorp on Wednesday January 25, 2023 @08:00PM (#34526452)
Yandex reverse image search is very good. I use it more than tineye, bing, or goog. It either gives you the exact matches if it can find them, or else it can infer what is desired and show many similar matches.
Re:Tell HN: Whole Yandex Git repository leaked by rcfox on Wednesday January 25, 2023 @08:29PM (#34526652)
I wonder if that's a function of their technology or their database. Google's image search used to be good, but regulations have forced them to cripple it in some ways, and they might have chosen not to surface results in some contexts.
Re:Tell HN: Whole Yandex Git repository leaked by ASalazarMX on Wednesday January 25, 2023 @09:06PM (#34526871)
Not only regulations, Chrome's builtin image search uses Google Lens to find products to sell you instead of simple reverse searches. It has become useless.
Re:Tell HN: Whole Yandex Git repository leaked by MonkeyMalarky on Wednesday January 25, 2023 @09:23PM (#34526996)
100% this. Google Lense is a product search engine, not image search.
Re:Tell HN: Whole Yandex Git repository leaked by recrof on Thursday January 26, 2023 @07:49AM (#34531013)
nitpick: "lense" is not a word.
Re:Tell HN: Whole Yandex Git repository leaked by mkl on Thursday January 26, 2023 @02:48AM (#34528877)
Agreed. You can currently do two-step process to get to the old useful search by clicking "Find image source" in the side-bar Lens results, but it's inconvenient enough that I barely use it any more.

In some browsers/profiles I use a "Search by Image" extension that still works properly.

When Chrome first made Lens the built-in image search, there was a way to turn it off. Does anyone know if that's still possible by some more-hidden method?

Re:Tell HN: Whole Yandex Git repository leaked by Psype on Thursday January 26, 2023 @09:57AM (#34532380)
Totally. Even the regular search is mostly made of selling links.

And in quantity, it seems like 9/10th of the content is gone.

Re:Tell HN: Whole Yandex Git repository leaked by essentia0 on Sunday January 29, 2023 @05:09AM (#34566666)
Agreed, nobody's life improved after they switched to that. There's stll a "Find image source" button above the picture in Google Lens that more or less does what you want
Re:Tell HN: Whole Yandex Git repository leaked by throwaway0x7E6 on Wednesday January 25, 2023 @09:38PM (#34527099)
>but regulations have forced them to cripple it in some ways

the regulations don't apply to Microsoft apparently, because even goddamn Bing has been better at it than Google for years now.

Re:Tell HN: Whole Yandex Git repository leaked by AlexanderTheGr8 on Wednesday January 25, 2023 @10:10PM (#34527282)
I never realized that reverse image search was so bad because of regulations. Can you give me some more context about which regulations affect google reverse image search and why?
Re:Tell HN: Whole Yandex Git repository leaked by rrdharan on Wednesday January 25, 2023 @11:04PM (#34527549)
I don’t know about reverse image search but I believe the previous poster had this in mind:

https://time.com/5163852/google-view-image-search-remove/

Re:Tell HN: Whole Yandex Git repository leaked by wolfgx on Thursday January 26, 2023 @01:12PM (#34535973)
Another story about Google image search when they removed "View Image" button just to end the feud with Getty Images.

https://www.digitaltrends.com/computing/google-alters-images...

Re:Tell HN: Whole Yandex Git repository leaked by unosama on Wednesday January 25, 2023 @11:20PM (#34527642)
It's not regulation. Google just has chosen to cripple it, the same way Twitter and Facebook mostly chose to censor certain viewpoints. All of these groups thing they are doing good.
Re:Tell HN: Whole Yandex Git repository leaked by trompetenaccoun on Thursday January 26, 2023 @04:32AM (#34529572)
That's part of the reason but it's likely also due to what user ogurechny explained. Most consumers have no idea about the true capabilities of their databases and it would scare them if they understood. Take facial recognition as an example. Also a non-crippled reverse image search would be the perfect stalking tool. Simply a lot of ways they could bring lawsuits and regulation on themselves so they tone it down just enough to not lose their search engine hegemony.
Re:Tell HN: Whole Yandex Git repository leaked by ogurechny on Thursday January 26, 2023 @01:27AM (#34528326)
I don't seem to recall any big story about regulation at the time. It seemed that they carelessly had their internal facial recognition tools (probably also used in Picasa, etc. grouping by person) run on all public content, and the results were feed into public image search. Then people started to notice that Google knows about ALL of their photos posted on the Web, and that probably negatively affected their opinion on the prospects of sharing all their photos and personal data with corporations.

Also, such data collection abilities are generally limited to governments, so it was clear that many of them (US first and foremost) would ask for exclusion of certain individuals, and so forth, and so on, so the public tools were crippled.

Yandex image search does have some facial recognition, but it also seems bit-starved and/or mixed with text search (there's a bigger chance to match if the name and surname is present).

Also, Google is pretty Victorian about porn these days. It's almost like it has a whitelist of “acceptable” porn sites to suit the tastes of potentially angry old ladies.

Re:Tell HN: Whole Yandex Git repository leaked by bakugo on Wednesday January 25, 2023 @08:30PM (#34526664)
It's also unexpectedly good at finding the original full images from cropped versions.
Re:Tell HN: Whole Yandex Git repository leaked by Strom on Wednesday January 25, 2023 @11:39PM (#34527727)
Not just cropped, the perspective and colors can be wrong. That is, you can take a photo of a poster on a wall from an angle, and it will give you the original image. It's impressive.
Re:Tell HN: Whole Yandex Git repository leaked by SanjayMehta on Wednesday January 25, 2023 @08:17PM (#34526574)
Yandex regular search is very good too, especially at surfacing results which appear to be censored by Google. The only problem is the interface language keeps flipping back to Russian.
Re:Tell HN: Whole Yandex Git repository leaked by kilroy123 on Wednesday January 25, 2023 @10:16PM (#34527308)
I agree. I use it when I'm searching for "iso" torrents.
Re:Tell HN: Whole Yandex Git repository leaked by account42 on Thursday January 26, 2023 @05:23AM (#34529953)
Is there a meta search engine that highlights results from Yandex that are missing from or ranked much lower on Google?
Re:Tell HN: Whole Yandex Git repository leaked by SanjayMehta on Thursday January 26, 2023 @11:01AM (#34533430)
Searx used to highlight the source engine. I don't remember ever seeing yandex though.
Re:Tell HN: Whole Yandex Git repository leaked by SV_BubbleTime on Wednesday January 25, 2023 @08:07PM (#34526499)
Hmm, didn't know they had one, but I'm let down by Google's 90% of the time so I'll try it next time.
Re:Tell HN: Whole Yandex Git repository leaked by ncpa-cpl on Wednesday January 25, 2023 @09:04PM (#34526857)
Their Street View product is also very good, however their coverage is mostly for ex Soviet Union countries
Re:Tell HN: Whole Yandex Git repository leaked by holografix on Wednesday January 25, 2023 @10:35PM (#34527398)
Had no idea this existed. Fascinating
Re:Tell HN: Whole Yandex Git repository leaked by Scoundreller on Wednesday January 25, 2023 @10:40PM (#34527420)
heh, I remember looking up some satellite maps of French prisons on Google maps. They were blurred!!!!

https://www.google.ca/maps/place/Centre+P%C3%A9nitentiaire+d...

A quick trip over to Yandex, and there they were in their full glory:

https://yandex.com/maps/10502/paris/?l=sat&ll=2.340173%2C48....

Re:Tell HN: Whole Yandex Git repository leaked by asdff on Thursday January 26, 2023 @04:11AM (#34529434)
It's funny when countries fuzz up their google maps images of military bases, or oil companies, or whatever. The US military clearly thinks doing that is a joke: you can zoom in on area 51 and look at parked aircraft.
Re:Tell HN: Whole Yandex Git repository leaked by astrange on Friday January 27, 2023 @08:41PM (#34554210)
The US just knows where all the satellites are well enough that they hide the good stuff inside when something's scheduled to fly over.
Re:Tell HN: Whole Yandex Git repository leaked by coolandsmartrr on Thursday January 26, 2023 @04:27AM (#34529537)
Probably because of legal compliance than technical reasons
Re:Tell HN: Whole Yandex Git repository leaked by account42 on Thursday January 26, 2023 @05:18AM (#34529906)
Good on Google for telling us what places to pay attention to ;)
Re:Tell HN: Whole Yandex Git repository leaked by 29athrowaway on Wednesday January 25, 2023 @10:00PM (#34527238)
It also sucks for privacy.
Re:Tell HN: Whole Yandex Git repository leaked by fIREpOK on Wednesday January 25, 2023 @10:04PM (#34527256)
Its good for privacy if you don't live in Russia... I don't think that they share the data with US governments?
Re:Tell HN: Whole Yandex Git repository leaked by timka on Wednesday January 25, 2023 @10:42PM (#34527431)
I tend to think that it's not the govs tracking us that we should care about but the big tech. Yandex is one of them.
Re:Tell HN: Whole Yandex Git repository leaked by fIREpOK on Wednesday January 25, 2023 @10:51PM (#34527479)
There are multiple levels of tracking... Governments being the top levels, then maybe global communication companies along with big tech titans... etc... AkA a complicated forest and/or tree of data?
Re:Tell HN: Whole Yandex Git repository leaked by timka on Thursday January 26, 2023 @05:31AM (#34530010)
That's complicated indeed. When I start thinking about who can track me (wi-fi, bluetooth and RFID scanners and surveillance cameras everywhere, my telecom provider, banks, public transportation, gmail etc., browser fingerprinting…) I just don't see any practical way to ultimately hide. Add the fundamental lack of trust to any device using ICs (evidence-based, field verifiable) and it becomes just insane.

In practice I just use a dumbphone and turn on my smartphone only to get a taxi (Yandex) or maybe see if my bus is coming (again Yandex). But my wife has her smartphone always on spying on us, not to mention all my colleagues. Also on my laptop I do use Yandex market, maps and read articles on Zen…

I can imagine that deep awareness of digital security issues may make one's life better in some situations but there's always a risk of becoming too much of an activist here. I think I'm quite geeky already.

Re:Tell HN: Whole Yandex Git repository leaked by fIREpOK on Thursday January 26, 2023 @11:39AM (#34534056)
> That's complicated indeed. When I start thinking about who can track me (wi-fi, bluetooth and RFID scanners and surveillance cameras everywhere, my telecom provider, banks, public transportation, gmail etc., browser fingerprinting…) I just don't see any practical way to ultimately hide. Add the fundamental lack of trust to any device using ICs (evidence-based, field verifiable) and it becomes just insane.

It is unfortunate that there appears to be only one way to reclaim our privacy,and that is through the passing of new laws and regulations. However, due to the way that laws are passed in the United States, I am not very optimistic. For instance, sometimes unrelated laws are attached to necessary budget bills in order for them to pass, or a three hundred-page bill of legal jargon is given a misleading title such as the "PATRIOT" Act (The unpatriotic bill). Furthermore, both political parties do not seem particularly supportive of privacy, if anything they appear to be opposed to it. We need to somehow make this an important political issue in 2024.

Re:Tell HN: Whole Yandex Git repository leaked by timka on Friday January 27, 2023 @11:23AM (#34547782)
Unfortunately it's impossible to fully elaborate my position here in the format of HN comments but I don't think the way you're talking about is really possible. The system is broken so badly that it cannot be fixed from inside by adding more of its rules. It has to be changed entirely. Figuratively speaking it's sort of like Goedel's theorem.

Have you had an opportunity to discuss their views on privacy and the currently established political culture in the US with those in their 70s and 80s?

Re:Tell HN: Whole Yandex Git repository leaked by xaerise on Thursday January 26, 2023 @07:32AM (#34530889)
And then we have China, where the big companies are required to share the data with the government if they request it.
Re:Tell HN: Whole Yandex Git repository leaked by fIREpOK on Thursday January 26, 2023 @11:24AM (#34533808)
> And then we have China, where the big companies are required to share the data with the government if they request it.

That's one thing that is probably the same in most countries... at least we know it is also true for USA and Russia.

Re:Tell HN: Whole Yandex Git repository leaked by soco on Thursday January 26, 2023 @04:28AM (#34529542)
Then the governments requesting from the big tech your infos (and getting them). Same thing in the end.
Re:Tell HN: Whole Yandex Git repository leaked by timka on Thursday January 26, 2023 @05:35AM (#34530038)
This is true. But 'the government' (I'd rather say agencies and services) mostly doesn't care about us unless we do something 'special' and you basically can't hide from them. The big tech is always interested.
Re:Tell HN: Whole Yandex Git repository leaked by ipaddr on Wednesday January 25, 2023 @11:49PM (#34527776)
Remember this yandex.com is hq in Europe. yandex.ru is hq in russia. Search carefully.
Re:Tell HN: Whole Yandex Git repository leaked by vbezhenar on Thursday January 26, 2023 @03:38AM (#34529226)
This is not the first leak from Yandex. So they "share" data with everyone. I used their Yandex food service. My personal info was eventually leaked and fraudsters started calling me. Not a huge issue but not a great situation either.
Re:Tell HN: Whole Yandex Git repository leaked by The_Colonel on Thursday January 26, 2023 @06:11AM (#34530273)
Eh, it shares data with Russian government.

If I have to choose between the two, I'll go with the US one.

Re:Tell HN: Whole Yandex Git repository leaked by not1ofU on Thursday January 26, 2023 @07:24AM (#34530840)
why?
Re:Tell HN: Whole Yandex Git repository leaked by The_Colonel on Thursday January 26, 2023 @07:41AM (#34530961)
Because the Russian government is much more dangerous to me than the US one.

Russia sent agents to blow up an ammunition depot ~15 kilometers from where I lived (google "Vrbetice 2014" if interested), killing 2 people. Russia regularly poisons and assassinates people. I mean, I'm no big fish to be interesting for Russia, but still the comparison comes out clear.

Re:Tell HN: Whole Yandex Git repository leaked by jeroenhd on Thursday January 26, 2023 @08:23AM (#34531310)
Russia is currently trying to invade and conquer even larger parts of Ukraine and a significant part of the world is helping Ukraine defend itself at the moment.

I prefer handing my data to privacy-breaching companies if the alternative would be handing data over to a state my government is currently holding a proxy war against.

Re:Tell HN: Whole Yandex Git repository leaked by Iwan-Zotow on Thursday January 26, 2023 @11:18AM (#34533715)
US right now invaded Syria and stealing Syrian oil
Re:Tell HN: Whole Yandex Git repository leaked by jeroenhd on Thursday January 26, 2023 @12:03PM (#34534549)
And if I were Syrian I certainly wouldn't want to use American services with the Russian government supporting the current regime.
Re:Tell HN: Whole Yandex Git repository leaked by fennecfoxy on Thursday January 26, 2023 @11:43AM (#34534152)
Lmao, bc Russians will murder me for being gay while the authorities look the other way, amongst all of the other atrocities that their state is perpetuating right now.
Re:Tell HN: Whole Yandex Git repository leaked by vintermann on Thursday January 26, 2023 @01:05AM (#34528197)
I think they also made it dumber after Bellingcat used it to identify a secret service officer. Interesting to see if we can find any trace of that in the git history.
Re:Tell HN: Whole Yandex Git repository leaked by computerfriend on Thursday January 26, 2023 @01:14AM (#34528242)
Allegedly it's snapshot data, not the actual Git repositories.
Re:Tell HN: Whole Yandex Git repository leaked by dotancohen on Thursday January 26, 2023 @01:24AM (#34528304)
By the size of the dump, my guess is that each tarball is a repo with the .git dir intact. But I cannot download it now.
Re:Tell HN: Whole Yandex Git repository leaked by spaceman_2020 on Thursday January 26, 2023 @02:50AM (#34528898)
Yandex translation and maps were also substantially better than Google in Russia when I traveled there in 2019
Re:Tell HN: Whole Yandex Git repository leaked by Finesse on Thursday January 26, 2023 @03:39AM (#34529234)
They are better not only in Russia. Many Google services feel outdated, sloppy and overcomplicated after using Yandex for a while.
Re:Tell HN: Whole Yandex Git repository leaked by actualwitch on Thursday January 26, 2023 @04:09AM (#34529415)
It is not hard being better at translation than google at this point. Probably the best option right now is deepl.
Re:Tell HN: Whole Yandex Git repository leaked by wolfgx on Thursday January 26, 2023 @01:15PM (#34536027)
I like that I can just drag & drop or paste any image into Yandex Translate website and it translates anything, also it has a better language detection than google translate, so you don't have to waste time to look the language you want to translate from. So far I haven't found any alternative, other than Bing translate desktop app...
Re:Tell HN: Whole Yandex Git repository leaked by tpm on Thursday January 26, 2023 @05:26PM (#34539773)
Google Lens translates images too.
Re:Tell HN: Whole Yandex Git repository leaked by NicoJuicy on Thursday January 26, 2023 @04:19AM (#34529475)
I actually doubt that a lot of that magic will be shown, as models ( because of ml ops) won't be in code.
Re:Tell HN: Whole Yandex Git repository leaked by comfypotato on Thursday January 26, 2023 @04:23AM (#34529505)
My understanding is that it has reverse facial recognition as well. Very cool. Also, it’s an interesting foray into the ethics of AI in that this could easily be used for nefarious reasons. (You can snap a photo of anyone who’s been indexed and find them online very easily.)
Re:Tell HN: Whole Yandex Git repository leaked by unkulunkulu on Thursday January 26, 2023 @05:48AM (#34530130)
Our cat once demolished a nice chair in a rental apartment. The chair belonged to the owner. At first we could not find where to get a replacement, but yandex reverse image search helped to my surprise!
Re:Tell HN: Whole Yandex Git repository leaked by wolfgx on Thursday January 26, 2023 @01:05PM (#34535833)
Agree. Here are two comparisons of some searches - [https://i.imgur.com/RmMWSG4.png] and [https://i.imgur.com/sIOhx6y.png] In 2022 even Bing beats Google..