Is this a defence?
It wasn't me, guv, it was my algorithm wot dunnit.
Google has been ordered by a German court to block defamatory words appearing in its search engine's autocomplete function. The Federal Court of Justice in Karlsruhe ruled that the advertising giant must remove libellous material from its algorithms that automatically attempt to guess what one is searching for - but only once …
If "Why is Lord McAlpine trending? *innocent face*" is to be considered libellous (http://www.independent.co.uk/news/uk/home-news/lord-mcalpine-defamation-case-only-a-moron-in-a-hurry-would-not-have-understood-sally-bercow-tweet-8575050.html) then typing "Lord McAlpine" into google and seeing "child molestor" pop up afterward seems be to an open and shut case of libel. It doesn't matter (as in poor Ms Bercow's case there) that she was only aggreggating and commenting on what lots of other people were commenting on, and nor should the fact that Google uses an algorithm that aggregates what lots of other people are searching on excuse it from blame.
"It was the machine wot dun it" is no defense, if someone programmed the machine to exhibit that behaviour.
If the algorithm is just based on the most common other searches then they are simply reporting a fact .
Here's a little thought experiment for you... perhaps someone might rent some time on a botnet of moderate size and generate a whole bunch of queries saying "Herr XXX interferen mit kiddies' and make Google's autocomplete start suggesting such things. Would you say that Herr XXX has been slandered in this situation? If so, do you think it is reasonable for Google to associate Herr XXX with kiddiefiddling, despite there being no such link in reality, on the grounds that they're just reporting facts about their search database?
On a semi-related note, I've just had an interesting idea about stock price manipulation. Fun times.
Interesting point however its not google associating Herr XXX with something nasty, its google reporting that other people are doing the associating.
If i tell my wife that a work colleague told me that his boss likes hairy dwarves and someone overhears am i liable for slander for reporting what someone else has said?
There could be an argument for slander charges against the bot net renter as they would be aware that they would be creating the association. Obviously that would be almost impossible to track.
Interesting point however its not google associating Herr XXX with something nasty, its google reporting that other people are doing the associating.
Is that sufficient to excuse them? As has already been mentioned, if I repeat some slander that some bloke down the pub uttered, even if I prefix it with 'I heard from some bloke down the pub' and suffix it with 'allegedly', I'm could fall afoul of libel and slander laws.
Because Google are allegedly repeating what other people have been apparently asking for, they can claim the usual safe-harbour, communications-carrier type defense, but in return they'd still have to remove the offending material. And that seems quite reasonable to me, though it might not please harder-core free speech advocates.
"We believe that Google should not be held liable for terms that appear in Autocomplete even though we created it, we manage it, we market it, we enable it, we embed it and we feature it on all of our web pages. Just because we are linked inexorably to a creation of our own design does not mean that we should actually be liable for it."
There, Google. Fixed it for you now.
Eadon I'm all for free speech but as Uncle Ben (the Spiderman version not the rice fella) says with power comes responsibility so how's about you become responsible and stop fecking derailing threads with your anti-MS agenda, its getting boring.
You don't like MS, we got that the 70th time you mentioned it so I think you can leave it alone now.
Beer because its one of those days when a cold one is necessary!
«"We believe that Google should not be held liable for terms that appear in Autocomplete as these are predicted by computer algorithms based on searches from previous users, not by Google itself," the company said.»
And the ruling isn't doing that; the algorithm can continue to do what it does. BUT if someone complains about a defamatory/libellous result from the algorithm, then Google must do something about, or they will become liable.
Exactly. Google arent being asked to change their results or their algorithms, they're just being asked to filter the autocomplete function. If someone still searches for the words together then there is no issue with what sites they find, but someone shouldn't immediately be branded as a scientoligists or a fraud just because google's algorithms say so.
But what is the barrier to complaining?
If Madonna or similar have a staff complaining about every "Madonna can't sing" "Madonna looks like an XXX" multiplied by 80million Germans it's going to have to basically block all names.
Then with libel tourism does it count if somebody in England types in the name of a German on the .com site?
"but someone shouldn't immediately be branded as a scientoligists or a fraud just because google's algorithms say so"
But that is not what is happening.
A friendly lady pops up and gives you a completely automated run-down based on a statistical survey. The fact that "$yourname suxx cocks" then happens to be in the result has no bearing on reality and cannot be construed as an "affirmation of fact" or a "branding". One might as well ban hidden markov models or the phonebook.
This can be usefully used, just google:
Was Elliott Abrams Hitler’s Senior Advisor?
We have two problems:
A screaming minority that is payed by the state and has too little work to do (No, not Wellfare recipients. I am talking about teachers and other "state employee for life"(1))
Courts that are a tad or three removed from the world
The majority of the germans do not care if the name of a state officials lady and "escort" pop up together. But this is simply another step towards the "D(eutsch) Netz" that will only include pages that are
+ Fully compliant with german laws as a Commisioner for Data Protection in Northern germany reads it
+ Fully compliant with the "ethics" as the german forum of Fosstard Research defines it (heise.de)
+ Matching the ideas of the Frank Freeloader Party
Granted, this net will consist of (very few) boring static pages. But it will be ethically, safe and protect your data
(1) Der Bundestag mal voller ist mal leerer aber immer voller Lehrer
Yes, we have two problems:
- You being a self-righteous egotist
- Making universal declaration on a group different than yourself because you simply must know better than they do
From playing the 'screaming minority lazy Socialists!' card to declaring what the average German citizen 'do[es] not care about', you've managed to hit all the "Fool" buttons with a single hand swipe. Good work!
Sorry, I realise it's OT and definitely Grammar Nazi territory, but I see it more and more and it annoys me in direct correlation with visibility. The phrase if "different FROM ..." , not "different than ...".
You differ from something, not differ than something
just like you can be similar to, or the same as, or equal to something.
You can also be greater than or less than something, but different than just doesn't make sense!
It's not defamation as it is a factual account of what people are searching for - no-one could be expected to believe that because an auto-complete shows text then that is a fact. However everyone is searching for that term for a reason and that is where the defamation lies, because someone has stated that "Bob is an Idiot" and a large enough audience has decided to find out more about Bob being an idiot.
As long as their algorithms are set to discount "auto-complete flooding" and that the terms only remain relevant for the period while it is still being searched for then censoring auto-completes seems a bit totalitarian.
However, I don't see Google winning this one and I would expect most countries to follow suit eventually to allow take-down requests of auto-completes.
Something to do with a guy searching his name and coming up with news articles about a sex offender (with the same name) or something of that ilk.
Here it is
and a more recent case here
Unless the man in question had such an unusual name that he was the only person in the entire world called that, then this really makes no sense.
If I Google my name, I can see people from all walks of life, the first auto-complete says I am a professor... which I am not, and nobody I know would think that that is me.
If there are other individuals who are Scientologists or have been convicted of fraud, and share the same name, then it is perfectly understandable that those terms would appear in a search. It is unavoidable. That anyone would automatically assume that those terms specifically applied to Mr X, is stretching things a little.
What the judge should have said was "The world does not revolve around you" and thrown this out.
Your idea is IMHO correct and actually that test apparently had been applied as the German press notice refers to a provision with that test. The reasoning is more or less as follows:
1. The auto complete refers to a person itself and not to a significant mass of people that can not be distinguished from the rest of the population -> Apparently yes in that case.
2. The auto complete contains a derogatory statement of fact that is not demonstrably true or a derogatory comment about a person (not its actions, its stated opinions etc.) -> Apparently yes, statement of fact assumed.
3. There are no mitigating factors (reasoning along the lines that who goes bowling has to bear that other people count the pins, e.g. "singer xyz can not sing") -> None given in the press release.
4. Google is responsible for this statement of fact -> Not in the first instance but when they were notified that there was something according to test 1 through 3, they kept on going and thus they are responsible from that moment onward.
Please note that the consequence according to the press release was only a cease and desist claim, not a damages claim.
So the press release basically states that prior to have knowledge of an infringement of personality rights, Google does not need to filter the auto complete results and they are not even required to exercise a standard of care to find an instance of infringement. Only when they were positively notified of the infringement, they have to act. This does not strike me as overly unfair but thank you for having brought up the point that you first need to run through tests 1 to 3.
Everybody either has one, is one or one enters similar search terms.
How can this be the fault of Google or it's algorithim when it is clear that the majority of people searching for these subject terms are looking for those specific words? Their search terms are effectively opinions, not libel.
Seems there are far too many absolutists in the German Government. Not everything in this world is either black or white.
However, an autocomplete algorithim that fails to look at the total number of similar searches will no longer provide meaningful results, thereby damaging Google. Or Apple, or Microsoft or whatever autocomplete algorithim is being pilloried today.
Cue cries of "Burn the Witch"
The needs of the many outweigh the needs of the one. (Especially when the "ones" are pederdasts or Scientologists).
Perhaps Google should provide a totaizer showing how many users seached with those exact terms, then there could be some idea of the reliability of results as well as proof they are not changing them.
"How can this be the fault of Google or it's algorithim when it is clear that the majority of people searching for these subject terms are looking for those specific words? Their search terms are effectively opinions, not libel."
Congratulations on supporting the court's very decision with your own words.
If "search terms are effectively opinions", then forwarding / repeating untrue to defamatory opinions in the eyes of the law is called libel. Google, therefore, has a legal obligation to make sure that transmission of defamatory opinions are controlled when asked to control them.
Because Google created the method of the transmission. That makes them legally liable in the eyes of the law. It is a lame excuse to say 'We have no control over the results', because you both created AND continue to administer the system that transmits. What Google is attempting to legally set as prescient is "once the computer system is born, we wash our hands of what it creates or outputs'.
Is that REALLY the prescient you want to legally construct?!
If so, you should carefully reconsider; it is NOT a good prescient. For example: if a government or company makes a system crawler that collects every single scrap of personal data on every single citizen to make a database, and then later intelligently follows leads and links into back door areas implied by the earlier data that it collected - isn't that creating a system and letting it built its own tables with no liability?
People on this board wish to deny responsibility for their code creations out of belief that, in a fashion, code is 'organic' - it does what it will do. That does not mean the world works that way; you are responsible for your creations. If your creations go and wander, doing some form of damage, it is completely unacceptable to state 'I am not responsible for my construct's actions'. Sorry, the rest of the world does not work that way.
Autocomplete was designed to build a table. The construct is both made and overseen by humans. That, in the eyes of the law, makes the humans ultimately responsible. I'm sorry if that legal idea hurts your personal ideas of code base being an entity onto itself. These are machines, we are their masters, and the buck stops with us because it has to - machines have no money.
Even though the complainant's name appears against "scientology" and "fraud" it's probably not him that is the reason for the appearance. Don't the German courts realise that many people have the same name. Even I with a very uncommon name have found that four other people in the whole of the UK have my name. If they were associated with something nefarious, should I sue them for bringing my name into disrepute or Google?
... it wasn't all that many months ago that Edzard Ernst was forced out of his post as Professor of Complementary Medicine at Exeter. He had had the timerity to assert that homeopathic treatments are biologically implausible and that clinical tests have shown they don't do anything at all in human beings.
Maybe people who buy expensive cosmetics and supplements are similar to those who go in for homeopathy. Maybe they mostly have more money than sense but prove to be useful 'friends in high places' when their suppliers are on the receiving end of flak.
I've become somewhat used to totally disagreeing with Google's point of view on the subject of just about everything these days.. However on this I find myself in total agreement.
An algorithm, in itself is neutral, inoffensive. The bits that are offending these plaintiffs are the words - and the words are user-generated.
So the moral of this story is if you don't want crappy things appearing about you (or your clients) on Google - try not to be associated with crap.
I'm pretty sure I can construct an algorithm that isn't neutral. But in the case of the methodology Google uses for their searches, I'm inclined to believe they are neutral. I'm also inclined to Google's side in this case, but more because it does involve futzing with the algorithm in a non-neutral way.
Having said that, I have to disagree with you on the last statement. It is possible to not be engaged in associating with crappy things and still have your name associated with it. Take Richard Jewell (RIP) for instance. His name is pretty much forever linked to Olympic Park bombing in a negative way even though he was truly a hero and did truly save lives.
Whether the German Court is right or wring to try to impose this ruling (I am not a lawyer), Google's response smacks of stupidity.
We know they already massage the *results* of the searches ("safe search" is a prime example) and do it very quickly and fairly successfully. But apparently filtering the *criteria* is too difficult?
or perhaps a better response would be for Google to remove autocomplete from .de and place a notice on the site saying "We have removed autocomplete due to a judgement in court X handed down by judge Y. We apologise for German users have less functionality available than other users as a result of that judge's decision" ?
Biting the hand that feeds IT © 1998–2019