Google turned me into a serial killer
hristo-georgiev.com
hristo-georgiev.com
To a non-technical user, I'm sure the box looks like a human-curated result which they're more likely to trust. Maybe that's the goal of Google's UI choices. Couldn't be further from the truth.
https://youtu.be/TbHBHhZOglw?t=58 (0:58 through 4:20)
it's terrible but a lot of people [0] think digital information, and thus internet, is truth
[0] and if google itself falls for this.. no wonder if people do too
Sarcasm aside, how do we propose they determine what is truth? If we assume the internet is full of information and more truthful than not, then Google’s assumption could be accurate. Of course they do try and solve this with the knowledge graph and expert curation. Connections to verified information might give validity to that information, but not always.
Google has been transitioning to attempting to provide a "truth engine" for several years. Whenever I try a complex key-word search, it suggests a question format for it (often with worse result but sometimes OK). When I have finally got the key words down to filter just what I want, google whines about "Not very many results, here's what you should do..." and, of course, Google often gives explicit answers for questions in it's search results (a notable percentage of which are wrong as noted).
And Google being half-assed truth engine is all sorts of bad...
The problem is in the presentation: Google's tools in general, not limited to search, present themselves as though they can identify truth. That's the flaw, the lie, if you prefer.
Google has started favoring info inputted into the Google My Business listing (the same one that dictates what appears in Google Maps for claimed businesses), so if the owners are updating their website but not updating that (or checking their email for emails from google that say '[business], are you open on 4th of July?') it'll show incorrect info.
This is the problem with the web search and scaping provider also trying to be a thousand other services at the same time. It's a constant conflict of interest.
Huh? If the business owners aren't updating the correct info in My Business, how can that possibly be Google's fault?
Google is a company that sells space for ads; an “advertising company” conventionally refers to a company that designs ad campaigns and purchases ad space for them on behalf of a client.
> not a search engine.
No, its both. Just as historically newspapers publishers in the US have been both producers of newspapers and companies that sell ad space.
If Google is scraping this information and keeping it separate, unlinked and un-updated from its source of truth, how is that anyone but Google's fault?
Once you claim your business location, most submitted changes should come to you for review. (There may be exceptions, I'm not an expert here)
Unwanted by whom? The reader? If you don't want google as an intermediary you can just... not use them. If you want to search but not have an intermediary, you can just run your own search engine and index. The publisher? Again, if a publisher doesn't want an search engine to be an intermediary then they can always get their site delisted. Turns out that most sites don't do that, or do and then quickly revert back, so clearly they want Google as an intermediary.
As the old adage goes. The only thing worse than no documentation is outdated/incorrect documentation.
Opening hours and Google Maps are the only things that get me on Google from DuckDuckGo these days (even if Google Maps keeps getting worse every year).
I think you have a decent chance of getting a five figure+ settlement from this. Talk to a lawyer about your options.
EDIT: When I search "Hristo Georgiev" (from US IP) there is no longer an image in the infobox. (As of 21:55:10 UTC, June 24, 2021)
I think a google engineer saw this HN post :-D
(You could still talk to a lawyer - remedying it now does not alter the fact that you were previously defamed. But Google has a stronger position having now remedied it)
OP appears to be in Switzerland, where I'm not familiar with the laws.
In the UK malicious intent is all that is needed - even if the statement is in fact true.
On the question of proving damages, that is the same in England and the US: both make a distinction between libel (or slander) per se and per quod. Libel per se covers statements that require no proof of damage because they are inherently damaging on the face of it. Falsely stating that someone is a serial killer who murdered five women is to say they are guilty of a crime of moral turpitude, one of the criteria recognised as actionable per se in both England and the USA.
In the USA, you have the issue of the application of the New York Times v Sullivan standard which grants First Amendment protection to potentially defamatory statements made about a public figure requiring proof of actual malice. Because of the press freedom protections contained in the First Amendment, statements about a public figure (or a limited purpose public figure) are granted a higher degree of protection. If a person is a public figure, then you would need to show that the person making the defamatory statement knew it was false, but if they are not a public figure, then that requirement does not apply. I think it rather unlikely that the present facts would lead to the Sullivan standard being applied: other than having a small personal website/blog to discuss programming issues, the OP is not a "public figure".
Regarding malicious intent, you don't need to show that in England. It helps if you can: malicious intent undermines a number of defences including honest opinion, and the publication on a matter of public interest defence (s4 Defamation Act 2013, and before that the Reynolds test), but it does not undermine truth as a defence. The burden of proving the truth of a defamatory statement does rest on the defendant in England in a way it may not in other jurisdictions. Malicious intent also goes to remedies. (It also wouldn't apply here: Google's algorithm cocking up is not "malicious intent", it is merely AI—automated incompetence.)
What the post you are replying to was likely referring to is the application of the "serious harm" test under s1 of the Defamation Act 2013, taken along with the requirement that the defamation amounted to "real and substantial tort" under the test established in Jameel v Dow Jones & Co. While I have considerable sympathy for the OP, the serious harm test is hard to find as you'd have to establish how many people actually saw the material and how many people were led to believe he is a serial killer rather than a person who happens to unluckily share the same name as a serial killer.
I wonder if defamation laws need to be extended to cover defamation by negligence?
Basically this meant you have to do research for everything you said. Could be a good thing, but this is not how the society works.
Against Google? I am more inclined to think the dude is actually a serial killer than that he has a chance of winning a defamation suit.
I think he has a very good chance of receiving a settlement, Google will not want to take this to court.
Did some googling (lol), was looking for any lawyers who have won successful defamation cases against Google. The top results that were successful suits in non-US jurisdictions (Australia, Japan, Ireland, Hong Kong).
Ex. https://www.mondaq.com/australia/libel-defamation/931462/goo...
I am not a lawyer, I just find law very interesting. I'd recommend reaching out to a few lawyers that deal with defamation claims in your current jurisdiction to see if they think you have a claim.
Direct link: https://news.ycombinator.com/item?id=27622100
Nobody was hurt, nobody suffered any loss, we all had a good laugh, and hopefully we all learned a lesson about what technology does, which responsibilities we all have in building better systems, and that humans, when building such systems, make mistakes. Nobody had malicious intent, on the contrary, the intent was to provide people with information when searching for it (whether in the name of selling ads or not is a different question).
Why do we need to pull lawyers and courts into this? Lawyers cost a lot of money and judges should not have to waste their time with this kind of thing. There are actual criminals out there, let lawyers and judges focus their energy on those.
P.S.: Don't get me wrong, I'm not defending that it happened. Of course it's bad, needs to be fixed, and be prevented for next time. But screaming "sue them!" just feels like dramatically overboard.
It is good for society when Google has to pay for their mistakes, because it encourages them to provide higher quality service and be more responsive. Just letting it go encourages them to keep cutting quality of service, because they can.
I believe that there is a simpler explanation.
The Wikipedia article is there in that side box because it is the top hit for "hristo georgiev" on Google's main search page. The picture is there because it is the top hit for "hristo georgiev" on Google's image search page.
The idea of mashing up the first image search result with the wikipedia snippet with no indication they are from totally unrelated sources seems pretty careless and irresponsible.
I seriously can't believe that the sources of the image and text aren't labeled with their respective sources. Doing so seems basic, obvious and trivial. Not doing so seems to be a blatant attempt to hide expected inaccuracies and make meaningless combinations of information seem more authoritative than it actually is.
While I think this individual would have a hard time in any court system, let alone the US court system, could Wikipedia perhaps have a claim of damages for libel (or something to this effect) due to misattributed information and reputational damage?
Or rather, all the knowledge graphic does is stuff not much more complicated than associate picture to name to article.
But what is pernicious is that presents itself as a knowledge graph and sometimes appears to have knowledge and so it seems to people to be a somewhat authoritative statement. And that causes people not-critically-thinking people to reach false and destructive beliefs.
I, being compulsively helpful, reported the error and it was quickly fixed. Maybe I'm part of the problem.
Amusingly that same infobox then has a table that states "2,560 stream processors" in the "Radeon RX 5700 XT" column. So one half of the ML parsed the web page correctly, presumably into some internal non-ambiguous representation, but then the other half of the ML interpreted it incorrectly anyway. Also, the "5500rx" (search query), "Rx 5700" (infobox title) and "RX 5700 XT" (infobox table header) are three different cards with different numbers of stream processors.
And people wonder why ML gets a bad rep.
The tech is pretty neat and all, but this is a blatant misuse.
- “Hristo Bogdanov Georgiev […] was a Bulgarian rapist and serial killer”
- “Cause of Death: Execution by hanging”
- “Died: August 28, 1980 (aged 23-24)”
Yet we also have
- Born: 1956 (age 65 years)
Also, if they don’t exactly know at what age he died, how do they know he would be 65 now, and not 64?
I think Google can do better, but not at the scale they need to cover all bases in Google search.
If you copied that correctly then that would imply that these results are personalized?
> a billion USD, Google could spend $100 worth of human curation per Wikipedia topic
$100 would buy how much? Probably 30 minutes of my time, if I'm feeling very generous. Actually, probably 10 mins if it's Google, because I'm not giving Google any discounts, if I'm to work for the Evil Empire, at least I want to get rich from it! That's not counting training costs, transaction costs, legal compliance and HR benefits costs, etc. etc. So, how much work I'd be able to do with one-time investment of 10 minutes? I don't think too much, even for the topic I'm an expert in. Maybe I'll be able to notice and fix one error, once.
And then, the data changes all the time. People do new things, people change jobs, people change names, people are born, people die. You have to run very fast to just stay in the same place. And then you have 200+ world languages (surprise, not everybody speaks English!) - Wikipedia actually has 300+ but let's drop the most exotic ones.
So a billion dollars wouldn't get you as much as you'd think. You probably need to bump it by couple of orders of magnitude. Which gives you an appreciation of how much value people are actually willing to donate completely free, if motivated correctly. Unfortunately, there's no way Google could have it - except through an intermediary like Wikipedia.
You're not the only one who can do the job, of course.
Hell of a feat, since he died in 1790.
> Rachel was a Biblical figure, the favorite of Jacob's two wives, and the mother of Joseph and Benjamin, two of the twelve progenitors of the tribes of Israel ...
... along with a happy smiling face of some office worker somewhere, named Rachel, of course. It was glorious.
Edit: I think the harder part would be showing that people would believe the claim (which legally is separate from showing harm). Google could argue that since the box showed the serial killer died in the 80's that a reasonable person would realize it must be a mistake.
A reasonable person would realize which part was a mistake: that it's a picture of the serial killer, or that the serial killer died in the 80s?
Suing in Bulgaria is likely to end up nowhere with Bulgaria having the worst courts in the EU (a primary reason not being in Schengen)
But damn was it satisfying.
There as been an interesting case - Aaron Greenspan vs Google: https://www.huffpost.com/entry/why-google-bothered-to-ap_b_2.... It's a very interesting and even entertaining read.
Let's start there first before issuing a lawsuit. You only sue if you can prove damages for defamation. Which, as tepid as most people are on here about patent trolls, I can't imaging taking on google. It's literally like taking on god at this point.
The only time I have seen actual people in the loop was on two rather large GCP users and even then it was sometimes just "have you tried restarting it?"-level of support half the time.
This reads like cheap, low-effort bashing.
It's like some guy punching me saying "For the Green Team!" and then you come by and punch me saying "For the Yellow Team!". Like, dude, you didn't undo the first punch. I'm now twice-punched. I want to be zero punched.
https://www.bbc.com/news/technology-47227937
Their automated processes do result in account deletions, and they intentionally build their systems so it is hard to reach humans to resolve complaints, so I don't think that comment was particularly undeserved - I read it as jokingly superlative.
Google screwing up their knowledge graph is neither "fake news" nor "cancel culture". Misusing these terms makes them useless for actual discussion.
I'm a decent read of people and talented at figuring out who actually makes sense and should be listened to. So going to people with domain knowledge and talking to them is usually the most efficient and effective means for me to get meaningful information when I am out of my depth.
It's also why I try to be patient with people online and answer seemingly "dumb" questions instead of telling people to google it. In many cases, if you aren't familiar with the subject, you won't know the best search terms and you won't know that the top result is commercial garbage and not really the gold standard source on the subject.
I routinely provide links for things like SRO because not only do people often not know that stands for Single Room Occupancy, if you google it you get a variety of unrelated hits (Standing Room Only, for example).
https://en.m.wikipedia.org/wiki/Single_room_occupancy
I was involved for a time with The TAG Project. I generally try to remember to provide a link for that as well because when I search for it, the thing I was involved with is not the top hit.
The top hit is thetagproject.com. I worked for tagfam.org and it is typically the third hit when I search for it.
I fairly often see people being obnoxious about "you should Google that" and I sometimes understand why they are aggravated with certain things, but I generally think that's asshole behavior.
"how many raccoons can fit"
No, I am not joking, and this is the result of 20+ years of web search development and decades of cutting edge AI research, right here.
Honestly, it pisses me off how bad Google's search is these days. It has close to zero clue about quality or relevance.
Basically, it's as relevant as searching for "the answer to life" and getting results spammed with 42.
Sorry, that is a bit over the top. What result would you expect for that query? Folks took a silly premise and made a joke about it and google in the absence of anything more relevant shows you their joke. While i was typing this query in I had to ignore the much more usefull suggestions: “How many raccoons in a litter”, “How many raccoons in the world”, “How many raccoons are in the us”, “How many raccoons live together”
And it provided reasonable looking lead to answer all of these question. And it did this while I was seriously mistyping the animal’s name!
If you don’t recognize how amazing this is then you left your blinders on. Let’s take this query for example: “How many raccoons are in the us”. This is a well formed human question, but it is not how one used to query a search engine. You were supposed to try to guess what words would appear on your imagined page and type those in. So for example you would type in “raccoon population us”. Except of course you were supposed to also know that the word “us” is ambigous, and appears too often in the wrong sense, so you would transform it to “United States” to help the machine. So by the expectations and conventions of the early google this is a badly formed query. A user error realky, yet now it can answer it! And it doesn’t just gives me a link where there might be an answer. Oh, no! It pulls the most important sentence out of the page and pasts it over the link.
This. Is. Freaking. Magic.
Are there mistakes? Sure. The linked serial killer thing is quite bad for example. But if you pick the raccoon example as your main argument then you lost me.
It really isn't.
I've been spending a lot of time doing home improvements and have often been frustrated by how difficult it is to get past marketing and spun content to find the information I need. Similarly I've struggled when researching some of the gnarlier aspects of leadership and the challenges that one has to deal with.
I'm sorry but whilst, yes, Google can perform some superficially impressive parlour tricks, it's simply not that great when you're looking for in-depth information, and it's particularly bad when you're looking for information that's not that far outside the mainstream but just enough so that what you need is buried in the midst of irrelevancies. Also frustrating when it keeps feeding you results that are from the "wrong" perspective (by which I mean not the perspective you're looking for, not that there's anything inherently wrong with the content being served up).
It is not magic at all: magic would yield better and more useful results.
Presumptuous of you to assume that's not exactly what I want to see when I search for that specific phrase.
"We made a profile of you, and if you think it's wrong, you'll have to register and share the right info with us" has been one of the safest giveaways of data hucksters. I used to think Google was the one exception, but by now I believe I should have trusted the rule.
At least someone is having fun with it: https://www.forgednfast.com/why-was-google-search-telling-pe....
This isn't a case where a highly uncommon name can lead to a high degree of certainty in association.
I can probably post all sorts of horrific crap on the internet without anyone ever being able to trace it back to me.
I went to grad school with someone who later lived in NYC. He shared a name with someone who basically was the cause of George Steinbrenner (NY Yankees owner) getting banned from baseball for a few years during the same period. Obviously not a popular NYC figure.
This was pre-web etc., but my friend ended up with death threats left on his answering machine.
Fortunately, none of us are too notorious, though one alternate "me" has pretty poor credit and another has been sued a couple of times recently.
I've said some time ago already that there is a multi billion niche waiting for whoever wants to do what Google used to do:
- input field in middle of page
- user types text into field
- software shows list of pages that contain said text. Modifiers can be used to influence exactly how exactly the matching will be
- the company is nice and reliable and goes out of their way not to be evil
This kind of thing should have very hard legal consequences for a company like Google.
Imagine being labeled as some kind of murder/rapist/pedophile whatever and moving into a neighborhood which gets angry fast.
"Just FYI, funny story, Google thinks I'm a serial killer. But that guy's been dead for years, and Google is mixed up."
Amusingly, I also get recommended a TEDx video titled "Hristo Georgiev: How to deceive Artificial Intelligence". Mission accomplished?
I do a lot of scientific image analysis using an ancient (but reliable!) piece of software called ImageJ [0]. There's a more recent distro of the same called FIJI [1]. So when I tried looking for how to extract EXIF data for GPS coordinates using ImageJ (not even mentioning FIJI), Google returned an info box about the Fiji-the-nation and provided the coordinates of said nation:
https://i.imgur.com/HxSh8Zv.png
In case anyone doubts that this is real I'm seeing it too https://i.imgur.com/PXepl6C.png
EDIT: Because they put an image with unterlated information together in such a way that it misleads people.
However, not all US states allow all types of damage claims and/or have special rules or higher burdens of proof related to those types of claims.
Generally speaking though, it is incorrect to say that somebody must show that they have had actual, monetary damages in order to be successful in a defamation lawsuit.
This overview from the Legal Information Institute (Cornell Law School) has some helpful info: https://www.law.cornell.edu/wex/defamation
You're basing your understanding of slander/libel on US laws I presume? If that person lives in Europe, generally, the bar for a successful lawsuit is __extremly low__ , it only requires the information to be blatantly false, there is no need to demonstrate the victim incurred any damages.
However if we are talking about monetary damages, those require hard proof of the damages and even then it is unlikely that they will have to pay all damages. Most lawyers I talk to usually recommend against suing for monetary damages.
Well, you can ask for "moral damages" and the judge might grant you these damages in Europe, I'm pretty sure the bar is also quite low for these. If you are from US keep in mind that million dollars damage verdicts are rather uncommon in Europe for individuals. That's the trade off. So judges might grant the suing party "moral damages" much more easily as a counter part. Of course it varies from country to country. IANAL.
edit: The article was an interesting read too.
In an ironic twist, the process of trying to get support from Google will probably drive him to become a serial killer.
That dataflow is human-curated.
I'm not searching a lot of Bulgarian names but I do search for npm packages and the like and despite my best efforts they frequently show me something completely different than what I searched for.
The saddest thing however is that DDG is just as bad, I use it just because I don't like Google and because it is easier to get from DDG to Google than the other way around.
> I started a project to unite all Google OneBoxes under this graph indexing system, which involved weather, flights, events, and so on.
Now we know who's to blame :)
Alternatively, add a drawing of the rapist to the original Wikipedia article.
Interestingly, for me, another Hristo (german principal investigator) appears on the right side when I google the name.
That Wikipedia article would probably meet the criteria for Speedy Deletion and just causes unnecessary effort for the Wikipedia editors.
"when was running invented"
"how many terashits does the ps5 have"
Running was invented in 1784 by Thomas Running when he tried to walk twice the same time”.
“69 terashits
ppl when the ps5 is revealed to have 69 terashits per megafart.”
What about all the others out there that can't do this for themselves?
Of course were I him it might be pretty bad. Serial killer is serious enough that people might consider there to have been a bug (as there was). But something less outlandish, like a misattributed fraud arrest, could have some pretty bad consequences.
I wonder how they do that. Do they just have a manual intervention list, where they can code exceptions to the ML results?
Otherwise it seems non trivial to quickly come up with another algorithm that does't have this particular problem.
17:19 Quick search shows your image has been removed from the right side panel.
Probably that's why :-)
Please don't kill me.
I wonder if the local sheriff could give him a small signed note explaining the problem to be shown to other policemen just in case. To assure at least that the local police in this place is aware of the situation (maybe ask them directly for advice?) could avoid future troubles.
The name is not unique - I personally know two people named that and in the whole country, there are probably hundreds. He has also stated that.
Just googled my own name. I know there are ~5 living people with the same semi-unusual name. They now show wikipedia text for an 18th century person (with the same name) along with a great 2020ish color photo of one of other currently living people with this name.
I've purposefully kept my few necessary online photos in grayscale, perhaps that helps ever so slightly with their brilliant industry leading AI algorithms...
When put like this, the thought that one company controls virtually all information flow on the Internet is more than mildly terrifying.
You're not wrong, but I think rather than bringing up counter-examples, you've produced a list of similar problems.
When someone tries to find a new website someone told them about for the first time... what % of that is filtered through Google? If it's a majority, which I suspect it is, then it's still definitely a problem.
The side effects of their search algorithms and how they control the flow of information has also been a topic of regular discussion/scrutiny for many years (just one example here [1])
> and if they really tried people would leave
Where would they go?
- [0] https://en.wikipedia.org/wiki/Censorship_by_Google
- [1] https://mindmatters.ai/2019/07/google-engineer-reveals-searc...
- Google doesn't censor anything
- Google censors things that a lot of people agree are offensive
The 2nd gets into tricky territory, because not everyone agrees on what is/is not offensive. I happen to think that most of the things they remove are probably good to remove. But the implication is still that Google wields (and at times uses) enormous power.
Things get even more grey (and potentially problematic) when you get into ranking algorithms, which have the power to sway opinions on major topics/events.
And as you mentioned, my parent comment is also a response to the very direct (and incorrect) claim:
> It seems that they do [control information], but they don't.
maps - apple, microsoft, openstreetmap
mail - dozens of providers
compute - amazon, microsoft, digital ocean, linode, rackspace
Not sure what else google do, I rarely use them
I am not sure about that. People are generally slow to change habits, especially that Google's control will not grow in a sudden fashion, but rather incrementally.
And let's not forget credentials. How many accounts depend on your GMail address? I don't know for you, but for me it encompasses most of the aspect of my life from banking to video games.
Which reminds me of this story: https://archive.is/EVrDv
> In the space of one hour, my entire digital life was destroyed. First my Google account was taken over, then deleted. Next my Twitter account was compromised, and used as a platform to broadcast racist and homophobic messages. And worst of all, my AppleID account was broken into, and my hackers used it to remotely erase all of the data on my iPhone, iPad, and MacBook.
github burned a house down twitter account I no longer have access to
Thanks google...
Suppose his name was "Thomas Edison"; would the title of the article then read, "How Google turned me into the inventor of the light bulb"?
Or suppose someone shared the name, with say, Slobodan Milosevic; is it Google's fault that a web search of that person's name turns up articles about someone who was charged with genocide and other war crimes?
I think it is mostly Google using his picture with the incorrect description below. I feel that is harder to catch for someone taking a quick glance than if the correct picture was used.