Google's Knowledge Graph info boxes are automatically-generated and littered with errors. I've been burned twice on operating hours, once for my local bank and once for a convenience store, driving over each time only to find the place closed. In both cases, the correct hours were posted on the business's website. I've also seen bad KG results for medical conditions, listing the wrong symptoms or describing easily-treatable maladies as "Incurable". Now I actively ignore the info box and intentionally click through to an authoritative non-Google website.
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.
Technology Connections had a great observation about this in his video about touch lamps: the problem with Google's knowledge graph is that it takes everything it reads on the internet at face value.
Google isn’t interested in truth, they are interested in information. They are provide a search engine not a truth engine.
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 isn’t interested in truth, they are interested in information. They are provide a search engine not a truth engine
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...
That doesn't need to be sarcasm, because it's true: at its core, Google's search is a method of finding information, not a method of directly ascertaining truth. It's not really possible for it to be a truth engine, and if you realize that, it's not even a flaw. You're left with a way of finding information you will need to evaluate for yourself, which is fine.
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.
Given that they are actively curating the information and censoring "misinformation", they certainly think they are a truth engine. And present it this way. Of course you'd only believe it if you believe Google is omniscient, omnipotent and benevolent.
That's the problem, isn't it. Most people want to be good, so most information on the internet is the-truth-as-they-know-it. It lulls you into a false sense of security.
if you peek outside of math/physics it's pretty much landscape of relative truths; imho ideal thruth/fake news detecting machine would simply require axiomatic/weighted input, ie. "I trust MIT with public key 0x..., youtube jesus from la with public key 0x... and my childhood mate 0x..." – based on that is "X true, false or undefined"? (weighted output). Because I trust MIT with weight ie. 500% and MIT trusts Caltech, my trust graph will favour Caltech view of the world. Yes you can throw blockchain and AI into it and it actually makes sense.
> . In both cases, the correct hours were posted on the business's website
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.
That's still Google's fault. If their service is providing out-of-date information, it's their fault. Making it sound like the business's fault for not using Google's other service is just reinforcing the idea that Google's information is the only information that's real.
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.
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.
Just because Google scrapes my website once to seed their database doesn't make me responsible for being their data monkey to keep it updated.
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?
You have to register your business page. It's not automatic. I believe op here meant a Google business listing which was created by the owner explicitly, but not updated.
It's not? Interesting. Literally hundreds of business listings popped up in Google Maps overnight, a few months ago, which appear very much scraped out of some registry - name, category, address, nothing else. Similarly with the """"place names"""" that seem to be based on people's personal remarks, and about as inaccurate.
Even if 90% of businesses play along (doing Google's data entry work for them), the 10% chance of my wasting half an hour makes it worth ignoring the info box and double-checking each time.
It has also been a waste of working hours at our offices. I work at libraries and am responsible for the communication towards visitors — we are slowly opening up all our venues again with lifting of corona measures, which mean we need to update our opening times. With one click I can reverse them on our own website but I have to update every location manually on google results. Anyone can also change these opening times plus they might just change them again based on some third party website that scrapes outdated opening times.
I don’t know what the prioritization is if you are e.g. explicitly registered in some Google database, but I’ve had a good experience with just marking up the opening hours (addresses, etc.) with RDFa (or Microformats or whatever else Google accepts these days) on the official website. Of course how easy it is depends on the number of layers between you and the actual HTML, and my case was rather straightforward (three locations with separately listed addresses and opening hours), but you might want to check if it’ll take some of the pain out of the process.
Google recently changed the hours of operation on my business' Google My Business listing to an hour later than I had it set to. I have no idea why; I certainly didn't tell them to. We found out when we got some angry calls from customers who showed up after we were closed. It's hard to explain to someone that we don't own our own listing on Google, and that we're at their mercy to approve the changes we submit and hope they don't make their own arbitrary updates.
The even uglier aspect of this, I think, is that Google is trying to keep you on their site and discourage you from visiting the authoritative site. That is classic middleman (unwanted intermediary) behaviour. Google produces no content but they sure interested in everyone else's.
> That is classic middleman (unwanted intermediary) behaviour.
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.
This is not true. While the content may initially be auto-generated, the Google KG is worked on extensively by real people. There are regional teams which work to verify and source the information that is out there. My sister works on the traditional Chinese team (mostly out of Taiwan or Taiwanese ABCs). These teams, as far as I know, are WFH contract workers.
Can users tell when info box answers have been auto-generated vs. curated? How often is KG content updated or re-verified, as in the case of businesses changing their operating hours? Are KG verifiers subject matter experts, e.g. medical professionals, or unskilled workers?
Real people or not, I've been burned by false opening hours on google maps so many times I will never trust it again. I'd rather spend 20 seconds extra to visit the official webpage rather than a 25% chance of traveling half an hour to realize it's closed. Even if extensive work by a human team can reduce this to a 10% chance it's still not worth my trouble.
As the old adage goes. The only thing worse than no documentation is outdated/incorrect documentation.
As far as I can tell the KG is also the source for Google Home when it answers questions. That's even more rough since it sounds authoritative and there are no visual cues to imply that it may be inferred.
Right. Voice assistants are a much harder format for Google, because you can’t just display a list of links. I ask my tube a question, I expect one (1) correct answer.
Google's medical information boxes are irresponsible. They are not accurate enough to serve the general population. They aren't detailed enough to be a useful reference for professionals either. Even a Wikipedia article is a better resource.
Opening hours has been mostly surprisingly accurate - but I agree, it failed a few times.
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).
You should seriously consider suing for defamation if this isn't fixed within a few days. This is an egregious error, especially if it was done by an automated system (and this could be a systemic issue affecting many others).
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)
In many countries you can't sue for defamation as easily as you appear to be able to the US. In the UK its required that you demonstrate loss or reputational damage, given how obvious it is that this is a mistake and the speed in which it was fixed, I think its highly unlikely to stand up in court. Simply calling someone a name isn't enough.
OP appears to be in Switzerland, where I'm not familiar with the laws.
You have it backwards, suing for defamation in the UK is considerably easier than in the US. The 1A in the US makes defamation and label very hard to prove - you must demonstrate that the statement is false, that the defamer knew it was false, and that you suffered damages as a result. All three must be met.
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.
"Google wrongly called me a serial killer" is a story many news outlets would love to run. He has the capacity to give them massive amounts of bad press, in a time where US regulators and politicians are closely looking at the dangers of automated algorithms from tech monopolies.
I think he has a very good chance of receiving a settlement, Google will not want to take this to court.
Author here - Given that I‘m not in the US puts me at a disadvantage. But if you can forward me a good lawyer who is willing to work with me on this, I may give it a shot.
I don't think that is necessarily a disadvantage - it could even work to your benefit as most of Google's legal muscle is in the US.
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).
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.
then the surprise twist during credits, while walking home after narrowly winning the suit against Google due to an impassioned statement on the stand, he murders someone!
I wonder if Wikipedia would have a cause of action here too; they're similarly shown to be falsely accusing a person via the inclusion of a photo which is not in their article.
Why is the top thread on HN about everybody suing everybody? Geez people, calm down.
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.
Google skimps on customer support because it is economically rational - the legal cost to them for the consequences of criminal or unethical acts perpetrated by their systems is much less than the cost of adequately staffing CS.
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.
> It turns out that Google's knowledge graph algorithm somehow falsely associated my photo with the Wikipedia article about the serial killer. Which is also surprisingly strange because my name isn't special or unique at all; there are literally hundreds of other people with my name, and despite of all that, my personal photo ended up being associated with a serial killer. I can't really explain to myself how this happened, but it's weird. In any case, I am now in the process of reporting this Knowledge Graph bug to Google.
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.
Yes, but: the text snippet is clearly labeled "Wikipedia" and is in the same box with the photo. Combined with the Wikipedia article being the first result, this would certainly give the average person casually searching the impression that the data in the box comes from Wikipedia, which tends to be a reasonably accurate, conservative source on living persons.
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.
One might be so inclined to label Google a peddler of fabricated misinformation...
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.
The info boxes sometimes contain surprising errors. I recently searched for Picasa and was informed that it was invented by Pablo Picasso in 2002! Google 'knows' Picasso died in the 1970s and it 'knows' he released a popular software program in 2002. That's an obvious contradiction requiring the simplest of rules to detect, but the system is just a dumb text extractor. (Interestingly, Google assistant gave the correct answer for 'who created Picasa?' so that system must use a different knowledge-base.)
I, being compulsively helpful, reported the error and it was quickly fixed. Maybe I'm part of the problem.
A few days ago, a person in an IRC channel I'm in was convinced AMD GPUs have CUDA cores, because they put "amd 5500rx cuda cores" into Google, saw the infobox entry that said "How many CUDA cores does a Rx 5700 have? 2,560 CUDA cores" and stubbornly refused to accept Google could be wrong.
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.
It's so strange, that infobox has never given me an accurate answer or an accurate question! What I mean by that is, it will find some snippet that sort of looks like it answers the question, but it's nearly always misleading. It's either lacking context, creating a question that doesn't make sense or just plain wrong. I avoid the thing like the plague and look for real answers in the actual search results (which are often the same pages, but at least provide context as to what question they are answering and how)
ML is not at fault - it's probably doing what it's trained to do. It's the marketing that makes people expect that if ML blackbox managed to do a couple of parlor tricks like distinguishing a cat from a banana and faking something resembling a human-written text, if you don't look too closely, then it's almost as good as a human brain. It's not. It's pretty much as dumb as a refrigerator (a very complex one, with many many buttons, but still a refrigerator). So you shouldn't expect too much from it, and then there's no reason to give it bad rep.
I think it's valuable to shift the focus from the information to the presentation of information. The information isn't controlled by Google, but the presentation absolutely is, and as mentioned elsewhere, makes the ML look authoritative, and seems to be implying an attribution to (if not outright falsely sourcing) Wikipedia, contradictions and all.
The tech is pretty neat and all, but this is a blatant misuse.
Interesting enough it tells me that he was executed by firing squad and not hanged. So do the screenshots of the source.
"Cause of death: Execution by shooting"
If you copied that correctly then that would imply that these results are personalized?
It's obvious for you, a human. There's no human curating those results (it's completely infeasible at Google's scales) and their graph building algorithms are extremely dumb. They are proprietary, but all of them are, so I'm sure Google's are too. The rules that could be added are obvious for each case, but there are trillions of possible cases, and there's no feasible way to create rules for even a small percent of them. So what you get is a giant mess of data that kinda sorta works most of the time, and spectacularly fails on any exception or complicated scenario. That's what we pay for doing such a massive information processing with such a relatively small amount of resources.
I’d push back on the idea that human curation is infeasable at Google’s scales. It might be more costly, but remember that Wikipedia did roughly the same task (accurate-enough text + image descriptions for every reasonable topic) with only volunteer time. There are under 10 million wiki pages, and nobody would claim English Wikipedia has any major uncovered areas. For a large-but-doable investment of, say, a billion USD, Google could spend $100 worth of human curation per Wikipedia topic. That represents a few minutes time for a skilled expert or a few hours time for an unskilled reviewer. Once you start refining your topic list, you can do even better than that.
Wikipedia relies on a vast army of unpaid volunteers. Google would never be able to buy as much time and effort, even with their trillions of dollars. That's why they are actively using Wikipedia (and Wikidata) data. And they also import errors and vandalism, and it takes much longer to fix errors than on Wiki, because see above.
> 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.
I think it was Rachel - many years ago, for a few days, when you search "Rachel" at Google it would show a snippet from Wikipedia:
> 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.
Sci-fi author Greg Egan has written about falling victim to a similar phenomenon, where photos of other people were showing up next to descriptions of him. No serial killers involved, though.
I'm glad this guy has a sense of humor about this, but I really hope that Google does right by him and that he doesn't get stuck in their Byzantine customer service process.
Falsely claiming that someone committed a serious crime is per se defamation in many jurisdictions. There is no need to prove actual damage. It is damaging by definition.
For libel, you generally have to show harm, but there are some claims that are assumed on their face to be damaging an no proof of harm is needed (libel per-se). Falsely accusing someone of being a serial killer would generally fall into this category.
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.
I don't think it matters whether they fix it or not. Evidence help with establishing the severity, but in most countries this sort of thing would be classed as defamation of character. This guy should be entitled to something and ought to sue regardless.
Easy to provide when will be detained at random airport for 10 hours until the authorities will triple-check all the info until everybody and the janitor will be satisfied. People has been stopped to take their fly for much less.
Might not be necessary. Many jurisdictions recognize certain types of claims, including accusations of serious crime, as defamation per se; no specific proof of harm is required.
It's surprisingly easy, just time-involved, to sue large corporations in small claims, at least in some jurisdictions. Yes, they've got lawyers, but if you have a case you may still win. I sued one of the largest telecoms in Canada for not honouring a verbal contract, and won quite easily once I proved the facts in dispute. Businesses assume people won't have the commitment to carry through and actually take them to court over malfeasance. Probably mostly correctly. I'm not sure it was really worth the hassle.
Out of curiosity...how did you prove a verbal contract? Recordings? 3rd person present as witness? Protocol from the verbal contract mailed to yourself?
Google is not a typical large corp though. They will spend millions on top lawyers for even small cases even if settling costs far less, just to make a point that they do not lose. If a company shows a willingness to settle or they they lose easily, then that makes them vulnerable to future litigation. Contingency lawyers will not take on companies that do not lose or settle.
You're replying to a comment that says "in small claims" and lawyers don't appear in small claims court. Though I don't know about winning vs Google there, just that it's worked against Uber.
I should have probably have noted that in most places, slander is not handled in small claims anyway. And lawyers can and do appear in small claims. If you sue a larger company, large enough to have a legal department, then someone from their legal department will probably show up to defend. You can consult or bring your own lawyer, too. A well-designed small claims court system keeps matters of law questions to a minimum, which usually makes it straightforward enough you can't drown it in paperwork. Here, there's a limit on how much you can spend on costs, for example. So no massive team of lawyers seeking depositions and hired expert witnesses. And as always, if it gets appealed, then who knows?
If Google refused to fix it, it might be different. But I expect that, if someone were to file a suit, not even trying to get it fixed wouldn't be seen as acting in good faith.
Does every person who is impacted by this kind of thing have the responsibility to do Google's QC for them? What about people who never notice it? Maybe a lawsuit would be hopeless, but it sure would be nice to see some kind of real consequences here.
How do you even contact Google about an issue like this in the first place? My only communication with Google in the capacity of a user has been automated messages.
I just looked it up and google does have a workflow that allows you to send them a court order.
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.
I found a Google One subscription is helpful for that. It's only a few dollars a month, which in fact I fund for free using Google Rewards dollars. You then get to speak to a real live person that actually respond properly, even with handwritten emails.
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.
Wait, how does that help? Low quality customer support hurts us. Low quality comments also hurt us. That just doubles the hurt for me. It doesn't compensate in any way.
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.
Google's (Youtube's specifically) automated processes have allowed the rise of extortion via their copyright systems using the threat of automatic account deletion.
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.
You missed the point. In a world of rampant “fake news” and “cancel culture”, Google screwing up their knowledge graph is dangerous ammunition that could be used to attack somebody’s reputation (which might lead to real/permanent damages before or regardless of whether the error can be corrected).
This kind of thing is part of why I go ask people questions in areas where I don't have a lot of domain knowledge rather than just search for it. (That's not to imply I don't do a search first. That seems awkwardly worded and I can't think of a better way to say it.)
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).
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.
For anyone who's curious but not enough to actually run the search themselves, the top results (using the string without quotes) all relate to how many raccoons can fit into a human anus.
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.
Well, apparently it's a meme. Yes it's objectively terrible, but I'll bet 80% of people searching for "how many raccoons can fit" are actually looking for those results. (I mean, unless you're buying a container to transport your twenty pet raccoons, it's just not the kind of phrase most people would spontaneously search for.)
Basically, it's as relevant as searching for "the answer to life" and getting results spammed with 42.
> Honestly, it pisses me off how bad Google's search is these days.
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.
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.
"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.
I have a longstanding project where I rank graduate creative writing programs by their alumni's appearances in a selection of prize anthologies. This means I spend a lot of time googling authors. There are a number of authors whose internet presence is shadowed by criminals with the same name. Then there are those who are shadowed by more famous people of the same name such as the Australian poet Kate Middleton or the New England essayist Ravi Shankar. Then there are the authors whose names are the same as other writers. So far I haven't had to do an IMDB-style (II) after someone's name although it's come close with some authors differing only by the presence or absence of a middle initial. And one instance I had to try three times to find the correct author of one particular name because there were two others (not anthologized) who published under identical names. I have a short story that turns on the whole name confusion thing that was published last year. https://sandyriverreview.com/wp-content/uploads/2020/11/2020...
I was so happy to find out 15 years ago that I have the same name as multiple pro athletes, both about the same age as me, too. I very much want to be un-Googleable.
I have a pretty common Dutch name. At my current company, I'm at least the second person with my first name-last name combination. Same at a previous company. When I became freelancer and registered with the chamber of commerce, a company with my exact name already existed.
I can probably post all sorts of horrific crap on the internet without anyone ever being able to trace it back to me.
The most problematic thing is when you share an unusual name with someone notorious who might plausibly be you at first glance.
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.
I have a relatively uncommon name, yet I've still come across two people who could be mistaken for me at first glance. One comes from the same county as me, and the other went to the same university. The former died 18 months ago in a car crash, so the first result for my name is now "[name] named as victim in [location] crash" (and there's a recommended search for "twitter" that uses my photo), which is a little strange to see.
I have an uncommon name yet there are at least 4 (I discovered a new one today via Google) of us who share the same name and are all within 2 years in age. One of the other versions of "me" lives in the same state I grew up in.
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'm actually not sure why I'm unique, at least in Internet times (someone much older turned up in a search at one point). Neither my given name nor surname are that rare. But they aren't super-common and are from different European nationalities so I guess that's enough to collapse into a unique point.
Which sometimes gives the false impression that criminals naturally go by all three names. A lot of TV shows will create a fictional serial killer named something like John Michael Doe because that sounds more "sinister" than just John Doe.
Its good to get this data point - from my familiarity with first names and last names, both Hristo and Georgiev are fairly unique and almost unheard of. The combination of both would qualify as a VERY unique name in my worldview.
But, in Bulgaria, eh, very common :-).
Is not really hard to imagine that more automation of this kind might result in some automated processes which results in someone get shot at a border by light handed policy.
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.
I mean, the fact that the serial killer in question has been dead for forty years makes this conclusively "a funny story" and not "a red flag". But it also probably is mandatory to cover this ice breaker before the person you swiped on Googles you.
"Just FYI, funny story, Google thinks I'm a serial killer. But that guy's been dead for years, and Google is mixed up."
I've seen how bad people are at skimming things they read online though even when the facts are laying there in front of them, and I could easily see someone (hiring manager, dating app they matched with) just googling their name, seeing the same pic match, and instantly running for the hills.
And, especially if it were something less extreme than serial killer and dates/location seemed plausible at first glance (or not--as you say skimming), a lot of people doing some quick resume triage will see the fraud conviction and move on.
For most people, if they came across a CV of someone who might be a serial killer, they'd want to read the full article. It's not something that happens every day.
Google your name from time to time. I do it to protect my personal information. I don't want my personal information, including my home address, email, phone number etc. to be exposed on search.
I literally just got my wallet back because someone Googled me, and could find an email easily since my site ranks well for my name so there's definitely some benefit to being exposed.
Having your personal email on your personal website is a little different from someone having posted your home address and phone # on some 'pay us money for people's personal info' website, though, and the latter is what you're going to spot by doing searches
Some things are public records or, like email, are pretty much inevitably exposed if you do things in public. But fortunately a lot of the "deep web" stuff that used to be free generally isn't any longer. (And cell phone numbers aren't accessible nearly as much as landlines were.)
I recently encountered the same underlying problem with Google's knowledge graph.
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:
Glorious. False positives on the synonym match gone wild. This is the kind of thing that deep learning was supposed to solve. The fact that it is failing from specific (imagej) to general (fiji) is beyond the wildest dreams of anyone doing quality assurance. I'm guessing the neural nets have been slacking by trying to do pure substitution matches and the devs have been rewarding them at the worst possible time. Wow. This is like seeing a toddler who is screaming in public and the parents are actively giving them candy to try and get them to stop. Wow.
This started going downhill about a decade ago. I wrote about it 8 years ago and I am fairly sure I had seen it going on for a while already at that time: https://techinorg.blogspot.com/2013/03/?m=0
Although in the US damages for defamation can include compensatory damages (intended to "make the plaintiff whole" by compensating for monetary losses) they can also include general damages for non-economic impacts (for example mental anguish & damage to reputation) as well as other types of damages.
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.
> Damages require proof. Unless they can show they lost something because of the mistake, then there are no damages.
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.
The barrier for a successful lawsuit is indeed very low, and based on my limited understanding, I think you will likely win when you demand that Google changes the picture on their website. They will likely also have to cover a part of your legal costs.
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.
> 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.
Wouldn't that be injuctive relief though instead of damages? Damages would be some kind of loss, which at this point there appears to be none as the author has stated.
Similar mismatch stuff exists with Google Scholar on two patent applications(now abandoned) where I contributed. It has my name but somebody else's photo and job title. Cannot apply to correct it just because I do not have any University email.
Google also has a problem with its news.google.com when they get their news from certain sources. Snoopes headlines are shortened and put up as if they were true. This resulted into some really vile headlines.
I was just throwing a number. I don’t work at Google, but I have heard from friends who work at that scale that when bugs only affect a few thousand people, it is “safe” to ship.
The error rate is so high that I see errors on a weekly or even daily basis when I fall back to Google and I only do 2 - 40 searches a day on Google these days.
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.
Suggestion: Create a new Wikipedia article with the same Name, upload your own profile photo onto it and put a Disambiguation (Programmer/Hacker) in it so that Google will associate it correctly.
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.
I added the wiki page for Kamala Harris' dad (a professor) to wikipedia (shortly after Biden announced her as his running mate in the election), and within literally 2 minutes, two different people had flagged it for deletion. One reason given was that academics need to be especially notable to warrant a wiki page, completely ignoring his relation to her daughter, which was mentioned in the original draft of the article stub. All proposals for deletion have since been removed, but I was surprised at the ferocity at which users want to delete new articles.
There's definitely a deletionist cult within Wikipedia. While acknowledging that it shouldn't be a dumping ground for essentially individual web pages, it goes too far IMO for people who have clear documented credentials and public history. This favors certain occupations over others of course--including academics, journalists, etc.--who have public bios, papers/articles, and often articles about them. But deletionist tendencies in these areas tend to work against claims that it's all about verifiability.
> Suggestion: Create a new Wikipedia article with the same Name, upload your own profile photo onto it and put a Disambiguation (Programmer/Hacker) in it so that Google will associate it correctly.
That Wikipedia article would probably meet the criteria for Speedy Deletion and just causes unnecessary effort for the Wikipedia editors.
If you find the right lawyer you will get at least $1 million out of this. Even if you don't feel like doing this, PLEASE do it for the greater good. Google will only start caring about these things if it costs them money. Money is the only language a corporation is fundamentally equipped to understand.
Next time LinkedIn asks why I don't have a profile picture I'm going to provide a link to this article. Scenarios like this are why I do my best to keep my photo off the internet.
You know what bugs me about these kinds of things - the millions of people that have already had this happen to them, but are terrible at public speaking or writing pretty much have no chance of getting these things corrected. Yes, we're all glad Hristo seems to have his situation fixed, but it did go viral and he's a great writer.
What about all the others out there that can't do this for themselves?
If I knew the author this would be cool: AFAIK I've never met a serial killer!
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.
This seems a hardcore case of moral damage worth at least some hundreds thousands dollars in a settlement or even suing some millions dollars out of them, doesn't it? Perhaps you should at least ask a lawyer before just humbly reporting the bug to them and forgetting.
Well, whoever googled him would conclude they should really avoid him. He could have missed a lot of great job, business and personal relations opportunities during the period Google was saying he was a serial killer.
Even if I didn't know about the mistake, I would still think idly that something was amiss, because that's the best-dressed and friendliest-looking goddamn serial killer I've ever seen!
I'm sharing a name and surname with a (Bulgarian)
criminal. Not sure what to do since it's not a problem so far and google doesn't show any of my pictures.
The riff on cancel culture sort of soured the entire post for me, to be honest. But otherwise I really enjoyed this read and hope you get it sorted out.
the best google result I ever saw was "disney children songs" with a google image search returning a very explicit pornographic picture.... I have no idea whether this is result of deliberate bombing or maybe something else. This does not surprise me at all - as I've seen completely wrong results for many other things...wrong years...wrong names...associated with people or events....
blows my mind that people didnt see this coming. and these guys want self driving cars. and all i ever wanted was a hoverboard that can go over water AND land. maybe fly an x wing. already it in VR. so im pretty ok with that. pretty disappointed that the future isnt more like the jetsons. it should go flintstones -> looneytunes -> jetsons
I hope he don't try to fly, or take a train, or cross a frontier in a few years, or things could escalate really fast. Some people wouldn't need much more excuses to shoot first and ask later.
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.
Bing has definitely not turned him into a serial killer. The main issue is associating his photo (wrongly) with the serial killer Wikipedia article. Bing is not doing that and instead showing a box with the Wikipedia article (without a photo, as it is in Wikipedia) and, separately, other results.
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.
Indeed, for a Bulgarian name it's about as far away from unique as it can be. Both the first name and last name would be in the top 10 most common names in Bulgaria. Imagine something like "David Brown".
This kind of mixup isn't too rare. I remember Googling "sigmoid" used to show a mixture of information about the function and the colon as if they were the same thing. There was also a famous case of something like "7 deadly sins" showing a rainbow flag for the sin of pride. It's creepy because it's presented in such a well designed integrated way yet a completely ridiculous mistake that a human wouldn't make.
Yeah, I've been googling people in the last months (job search, interview prep) and google is really bad with homonyms. Occurrences like in this article is the norm when you're not looking for someone famous. Says a lot that Google rolled it out.
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...
It seems that they do, but they don't. Plenty of information flows through emails, chat, social media, direct search on websites like Amazon. The internet is large, and Google is a very big player, but they're not controlling all the information flow on the internet, and if they really tried people would leave.
How much email flows through gmail? Should Amazon have 50% of all ecommerce transaction data? How much of the chat space and social media does Facebook own?
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.
That Google controls/censors information is well established [0].
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])
I know you're trying to make a point about censorship existing, but I'm not sure if loli porn and scam online pharmacies were really contributing to the internet.
I agree with you; however, that's kind of the point. Google has the power to organize and control a significant amount of information. These statements are very different:
- 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.
> they're not controlling all the information flow on the internet, and if they really tried people would leave.
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.
Well yeah, but you now have listed basicly 3 companies with most of them having a majority market share in their domain. (Gmail, FB, FB, Amazon; respectively) Plus google search is the end-all-be-all answer. People treat these results as the truth.
Luckily for me, my name is the same as a formerly popular folk musician who lived near my hometown and made songs whose titles had my hometown's name in it.
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.
> 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.
Board members don't have the direct control that you think. Neither do all the C-level officers. It's a basic fact of organizations that people that make the key decisions aren't necessarily the ones with the titles.
It boggles my mind how we went from a google search result to white supremacy.
I'd like people to please exercise restraint, the world feels like it is falling apart where every conversation becomes political/racial. If this is going on HN, I wonder how it is elsewhere.
Yeah, it’s a bit much. The “extreme” reading is slowly becoming the only acceptable one. Partly because of comments like the one precipitating the thread: there’s no good reason to try to connect these items.
The 'white males are the enemy of the people' is such a leftist trope that it's almost hard for me to believe someone thinks it's an alt right dog whistle.
It's the old correlation/causation problem. Being white is at the present time in history correlated wih privilege. But being white isn't the problem. The problem is how the privileged use it to lock out everyone else from the benefits they enjoy. Race and such like are irrelevant distractions.
Here's some mic spam I heard the other day when gaming:
"There's all sorts of N*s - light brown, dark brown, even white ones. ...<other examples>... and even the homeless guy on the corner is a N"
It's not the first time I've heard this sentiment, and presumably not the last.
Perhaps race (and gender, sexuality, wealth, etc) are related to privilege. I mean they are used to justify why entire groups of people deserve more or less privilege. Maybe those things are aren't "just a confusing distraction" but their associated 'isms' drive the issue of privilege.
Or maybe they are divisions used to distract from the privileges as you say.
Either way, dismissing race as just a distraction, that is to treat different facets of a bigger problem as unrelated, seems like an attempt to quash any discussion of the whole.
> ...he can basically be considered white for purposes of discussing his privilege.
This is a distraction, and a confusing one. Calling that out is not an "attempt to quash any discussion". Let's discuss race when race is relevant, and the other 90% of the time let's discuss economic privilege.
I started my carreer with no debt for education. My family isn't rich but they aren't poor - just somewhere in the middle - above median income but not in the top tax bracket. They were able to provide help and I worked a bit to make up the difference. It's a pretty good leg-up for me, and I'm glad for it - it has made my adulthood much better than if I had to pay for loans and such.
This can directly be traced to the fact that my under-qualified father had a good job and worked hard. How did he have a good job if he was under-qualified? Well the labor pool was artificially kept small by disallowing women and people of color, and he went to college without debt. Why did he get to go to college without debt? Well his dad was able to own properties in neighborhoods where he could charge good rent, and he kept those properties nice. He was allowed to do this because he wasn't black and thus could legally own those properties, again artificially reducing his competition. He worked hard at keeping those properties nice, but he also hired a lot of people of color for low wages - they also worked hard.
So did my position happen in life because of economic privilege that my family built? Did my position in life happen because I'm white? The answer to both is the same: In part.
They worked hard. They spent wisely. They also didn't have to worry about a lot of potential competition because they weren't limited by racist rules and laws. Would they have done as well if they didn't get those protections? I doubt it though - they are good people but not top tier at thier chosen professions.
My point is, these things are really mixed together - my story here is pretty common - there's no denying that economic privilege I've enjoyed is in part because I come from a family with a bit of economic privilege - privilege gained in part due to their race.
It's better today, but it's not "fixed" by any stretch. There are still realtors that get in trouble for redlining or rejecting rental applicants for their race. There's plenty of examples, from this year even, of the same home being appraised at different values depending on if the photos around the house were of black or white families.
When something as fundamental to economic privilege as "ability to purchase things I can afford" and "ability to sell at market rates" are affected by race like that, is it really possible to say race isn't relevant to economic privilege?
The thread began when it was pointed out that it wasn't just any corporation to which we handed so much power over us, it was a corporation dominated by white men.
The reason I believe this is unhelpful, confusing, and irrelevant, is that I believe that Google is actually one of the least racist powerful institutions on Earth, throughout both time and space. Don't get me wrong, I hate Google as much as any other ad-powered information harvesting giant, but they're a company that literally won't shut up about Black voices and women in tech. Which, more power to them for that! I think it's basically cynical from an executive's perspective, but cynical people can do good, so whatever.
But anyway, the point is I don't think race is relevant at all in the conversation about how Google has too much power. They use that power for evil quite often, but when they use it for good, the good is often trying to solve racism. Is it really helpful to point out how white they are?
If you search my first and last name, the first result is of two 18 year olds who burned a house down because one kid has my first name and the other has my last name. For me it goes:
github
burned a house down
twitter account I no longer have access to
I don't see what the big deal here. The serial killer really did exist, and has the same name as the author of the blog post. Searching for "Hristo Georgiev" using Bing and DuckDuckGo turns up the article about the serial killer as the first link. Why is this not "DuckDuckGo turned me into a serial killer", other than people hating on 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?
The whole point of the post is that Google grabbed his picture and attached it to the search card of said killer, so it looks like he is that person not only by name, but by first searched photo as well.
>Why is this not "DuckDuckGo turned me into a serial killer", other than people hating on Google?
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.
the big deal is that the author's photo is shown next to the information about the serial killer. If that wasn't the case, it wouldn't have been a problem of course.
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.