Become a fan of Slashdot on Facebook

 


Forgot your password?
Close
typodupeerror
×
AI Technology

AI Still Useless at Catching Hate Speech, Research Finds (theregister.co.uk) 238

Posted by msmash from the letting-humanity-down dept.
New research has shown just how bad AI is at dealing with online trolls. From a report: Such systems struggle to automatically flag nudity and violence, don't understand text well enough to shoot down fake news and aren't effective at detecting abusive comments from trolls hiding behind their keyboards. A group of researchers from Aalto University and the University of Padua found this out when they tested seven state-of-the-art models used to detect hate speech. All of them failed to recognize foul language when subtle changes were made, according to a paper [PDF] on arXiv. Adversarial examples can be created automatically by using algorithms to misspell certain words, swap characters for numbers or add random spaces between words or attach innocuous words such as 'love' in sentences.
This discussion has been archived. No new comments can be posted.

AI Still Useless at Catching Hate Speech, Research Finds More

AI Still Useless at Catching Hate Speech, Research Finds

Comments Filter:

  • No such thing as "hate speech" (Score:5, Insightful)

    by Alypius ( 3606369 ) on Friday August 31, 2018 @01:02PM (#57232114)
    How do you catch something that doesn't exist?

    • Re: (Score:2, Funny)

      by Anonymous Coward

      How do you catch something that doesn't exist?

      Trained snipe and jackalope(s) sniff it out.

    • Unless there are "happy murders" and "love frauds" ?

      • Re: (Score:3, Interesting)

        by blindseer ( 891256 )

        Unless there are "happy murders" and "love frauds" ?

        Murder is by definition unlawful. Homicide on the other hand has four kinds, felonious, excusable, justifiable, and praiseworthy. At least according to Ambrose Bierce. That might equate to (by my estimation at least) murder, manslaughter, self defense, and war.

        I remember a short exchange on the distinction between killing and murder in the movie The Big Red One. In short is that you don't "murder" a Nazi, you kill them, much like one would kill a rabid dog. A sick dog isn't ever murdered because there

    • Re:No such thing as "hate speech" (Score:4, Insightful)

      by alvinrod ( 889928 ) on Friday August 31, 2018 @01:33PM (#57232428)
      I don't think there are many people (even among strong free speech proponents) who will deny that hate speech (or perhaps hateful speech if one wanted to get truly technical) exists. Rather the position is that even though it may be hateful, offensive, or otherwise displeasing to some people, there is still no reason for the government to prohibit someone from uttering it.

      Free speech (and the desire to protect it) is necessary precisely because there are things like hate speech.

    • Oh it does exist. The problem is that it is wrapped in often a complex set of explanations, that makes it difficult for a computer to figure out if it is hate speech or just a deep explanation of the problem.
      The core of the problem is people spend way too much time and effort to validate being cruel to other people. Discrimination and Hate speech compared to explaining racial and cultural difference often difficult.
      Usually the key difference is the result of the premise, which is this is why X groups is

      • No, the real problem is hate speech is totally subjective. People can find the exact same statement innocuous or hateful so it is no shock that computers can't figure it out as we can't figure it out. Hell, just look at the most recent 'monkey it up' controversy. Many see this as a highly racist and hateful statement. Others see no racism in it at all. And this why people say it doesn't exist as they say there is no objective standard for such, and there isn't. Now, I don't think that logic follows, b

      • It is also often far more "hate speech" today on the far political left than on the far political right, against whom the statutes are enforced. It's far easier to charge a skinhead than a 6 foot former football player transgender person who grips the back of the neck of a debate opponent who is at least 5 inches shorter and 70 pounds lighter and promise to send them home in an ambulance.

        https://www.youtube.com/watch?... [youtube.com]

        It's a very, very

    • Hate Speech is speech specifically intended to incite, encourage or justify hatred; often racial hatred but also includes religious, class, and a variety of other hatred. The key feature is that hatred is directed at a specific group of individuals because of those individuals membership in that group and not because of any specific actions by those individuals or even that group as a whole.

      It is commonly used by a ruling class in order to divide the working class into manageable chunks. It is part of a

    • The algorithm is very simple:
      1. If speaker=conservative, -> hate speech
      2. If statement != progressive -> hate speech
      3. If 1)+2) -> nazi

  • On the other hand... (Score:4, Funny)

    by JoeDuncan ( 874519 ) on Friday August 31, 2018 @01:04PM (#57232138)

    ... they're fucking *brilliant* at CREATING it!

    https://www.theverge.com/2016/... [theverge.com]

  • And at almost everything else (Score:3)

    by OneHundredAndTen ( 1523865 ) on Friday August 31, 2018 @01:07PM (#57232162)
    The current AI offerings (Google Assistant (or whatever it's called today), Alexa, Siri) remain extremely limited in what they can usefully do. Try something with even a minimum of ambiguity, and they start spinning their wheels real fast. Even when dealing with very simple queries, it is obvious that their claims to intelligence are laughable. One of my favorite examples: "Ok, Google (or Alexa, or whatever) do not, under any circumstances, give me the weather forecast". Sure enough, they all promptly will give you the weather forecast. The truth is that, as of today, most of the time it is easier, faster and more efficient to do the job yourself, rather than trying to coerce them into doing it for you.

    • The truth is that, as of today, most of the time it is easier, faster and more efficient to do the job yourself

      I agree. Not getting the weather forecast is much easier if you do it yourself.

  • The sentence is complete and should be stopped after the 1st 3 words.
    "AI still useless"...
    why? AI doesn't exist yet.

    • AI doesn't exist yet.

      General, human level AI doesn't exist yet, but that's only a small subset of the field. AI is actually pretty ubiquitous. There' very likely a fuzz logic controller in your dishwasher and washing machine to control water levels.

    • Hear, hear!

    • Re: (Score:2)

      by Megol ( 3135005 )

      Depends on how your define AI. Some consider expert systems AI and, well, those have existed for a long time.
      There aren't even good general definitions of the word intelligence.

  • AI-ish. (Score:5, Insightful)

    by devslash0 ( 4203435 ) on Friday August 31, 2018 @01:12PM (#57232214)
    That's because all of the learning/reasoning software that we proudly call AI is not AI at all. It's just a series of pattern recognition and reasoning operations which are only as good as the programmers behind it. Some systems are okay. Most, however, terrible. Even if we got anywhere close to true AI, it would have to be more sophisticated than our intelligence to outsmart human deception. It would need to know about not only our language but also psychology and in fact everything we've ever created to have the full context of each end every conversation.

    In other words, hate speech filters are not likely to start working any time soon.
    • You're the second person to comment who sees this, and you're completely correct. The current excuse for 'AI' lacks any actual cognitive capability -- because we can't write software to emulate something a biological brain can do, but that we don't understand how it does it. Too many people, I believe, see TV shows and movies with totally fictional 'AI' in it, and they think it's the same thing. Not even close.

    • There are plenty of AI systems that are better at performing tasks than their programmers could do.

      • performing tasks is a somewhat finite subset of life experience and begin good at repetitive execution does not constitute full machine intelligence. AI has been touted as being just about ready, continually for a half century, yet we can see that it is still unable to grasp context, nuance, intention, or any of the myriad emotional motivations that comprise random living beings.

    • Re: (Score:2, Informative)

      by JoeDuncan ( 874519 )

      That's because all of the learning/reasoning software that we proudly call AI is not AI at all.

      Yes it is.

      It's just a series of pattern recognition and reasoning operations ...

      .

      Uhm... but that would make those systems AI and you just said they weren't, so now you're pulling a "Trump" and contradicting yourself within a single statement.

      ...true AI...

      Ah! I see the confusion! You don't know what the fuck you're talking about.

      By "true" AI, I guess you mean "general AI" of human intelligence? Actual AI encompasses a LOT more than that.

      In other words, hate speech filters are not likely to start working any time soon.

      • but that would make those systems AI

        Oh, please. I've written "pattern recognition and reasoning operations" for my entire career. Using that as a definition is too vague to be of any use.

    • No, that's not why (Score:4, Insightful)

      by SuperKendall ( 25149 ) on Friday August 31, 2018 @01:49PM (#57232566)

      It's just a series of pattern recognition and reasoning operations which are only as good as the programmers

      All that is true but it's not the fundamental reason for failure.

      The fundamental reason for failure is, there is no such thing as hate speech. You cannot write some reasoned algorithmic or pattern based approach to detecting something that is only detectable by someone with hatred inside them.

      Until we teach a true AI to hate, it in turn will not recognize speech that drives it mad and thus should be declared "hate" speech.

      • The term hate speech was invented in the nascent PC culture as a convenient tool to, first ignore and then attack other persons with whom one has a disagreement. It was extended to hate crimes in order to incrementally increase punishment in a vain attempt to correct peoples' thinking when it doesn't conform to one's own perception of political correctness. Thus it is totally subjective and fluid, so that any attempt to codify it as a set of AI rules is bound to fail.

      • ...there is no such thing as hate speech...

        So your saying it's literally IMPOSSIBLE to advocate for physical violence against an identifiable group of people?

        Like, if I wanted to, I would be unable to utter the words: "Let's go kill all the republicans"?

        • So your saying it's literally IMPOSSIBLE to advocate for physical violence against an identifiable group of people?

          I am saying that is not hate speech. That's called "shit talking". Or maybe it truly is a call to violence. Depending on how written maybe they should be having a police visit.

          It could also be sarcasm. It could be anything.

          But "hate speech" has no meaning as a term. Some people are claiming even non-speech things are hate speech, that is how weak and pointless the term has become.

        • ...there is no such thing as hate speech...

          So your saying it's literally IMPOSSIBLE to advocate for physical violence against an identifiable group of people?

          Like, if I wanted to, I would be unable to utter the words: "Let's go kill all the republicans"?

          You're conflating "hate speech" with "incitement". They are two separate things. "Incitement" as in "incitement to violence" is a pretty clear and objective standard. "Hate speech" is anything but objective or clear and changes depending on the "victimhood status" of the particular speaker.

          The resurgence in Post-Modernist thought is the genesis for "hate speech", "safe spaces", "white privilege", "micro-aggressions" and many other of the relatively recent social bu

  • I don't need your protection. (Score:5, Insightful)

    by Harvey Manfrenjenson ( 1610637 ) on Friday August 31, 2018 @01:17PM (#57232262)

    I've somehow managed to get by for many years without any algorithms to "protect" me from speech. So here's my message to the inventors of this algorithm: I'd love it if you would fuck right off, OK? Thanks and have a great day!

    • Re:I don't need your protection. (Score:4, Funny)

      by Harvey Manfrenjenson ( 1610637 ) on Friday August 31, 2018 @01:30PM (#57232392)

      Amused to find that within two minutes of posting, my contribution was modded "50% Insightful, 50% Redundant".

      I have to admit, it's a fair enough grade. Even as I posted I was thinking to myself-- what I am saying is so blindingly obvious, does it really need to be said at all?

      But I think the very existence of this "research" shows that it does need to be said, loudly, and often, and by as many people as possible.

      • There isn't a single algorithm at play, it is all the algorithms that are being generated by morally unmoored young engineers. Recall when Google was in its infancy, with the heady idealism that everyone is innately good, so they used the mantra Don't Be Evil and recently abandoned that when it began to hinder profit opportunity. At the outset they tried to be moral, yet what anchor were they using? Simple idealism failed, so now there is no anchor, and we can see it in FB which was designed as an incel st

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      It's not for you, it's for Facebook. Unsupported Facebook doesn't want certain content on its platform, e.g. illegal images.

      Policing content is very labour intensive, so they create tools to reduce the burden. Blocking known images, for example. But that still means a lot of human review, which is not only labour intensive but pretty hard on the reviewers too.

      It's even worse with hate speech. They can do some simple pattern matching like "1488" but most of it is users reporting material and well beyond the

      • It's not for you, it's for Facebook. Unsupported Facebook doesn't want certain content on its platform, e.g. illegal images.

        Policing content is very labour intensive, so they create tools to reduce the burden. Blocking known images, for example. But that still means a lot of human review, which is not only labour intensive but pretty hard on the reviewers too.

        It's even worse with hate speech. They can do some simple pattern matching like "1488" but most of it is users reporting material and well beyond the ability of AI to evaluate.

        Which is why the *only* solution is to pass a law so that Facebook, Twitter, ISPs, etc are not held legally liable for the content other people disseminate on their platform.

        These entities are not "publishers" in the traditional sense. They don't read a post before it goes live; they exercise no editorial control prior to publication, and as you point out, it would be humanly impossible for them to do so (because of the vast amount of content being published every second). They are more akin to a phone co

        • Re: (Score:2)

          by AmiMoJo ( 196126 )

          It's not a question of legal liability. Facebook doesn't want certain legal content on its service because it wants to be an attractive destination for the majority of users.

        • Re: (Score:2)

          by Raenex ( 947668 )

          Which is why the *only* solution is to pass a law so that Facebook, Twitter, ISPs, etc are not held legally liable for the content other people disseminate on their platform.

          Are you being sarcastic, or do you not know that they already enjoy this legal protection?

          • Which is why the *only* solution is to pass a law so that Facebook, Twitter, ISPs, etc are not held legally liable for the content other people disseminate on their platform.

            Are you being sarcastic, or do you not know that they already enjoy this legal protection?

            It's more complicated than that. See for example: https://www.npr.org/sections/a... [npr.org]

            Really, it's more complicated than my earlier post suggested. The issue is not merely whether Facebook is protected *now*: it's whether they can count on that protection going forward. As the recent congressional hearings with Zuckerberg made clear, Facebook is terrified of what laws Congress might come up with to punish them, and they are currently scrambling to take steps to *prevent* that happening. Hence the renewed i

            • Re: (Score:2)

              by Raenex ( 947668 )

              It's more complicated than that. See for example: https://www.npr.org/sections/a... [npr.org]

              I had forgotten about that, but it's really limited in scope and that's not what is pushing the current censorship. What's really going on is that the Big Tech oligarchy leans heavily left, and they ramped up their censorship to disparately impact conservatives right before the midterms. And if they're going to be political about their censorship, I don't think they deserve legal protection under the law for their content.

  • False positive (Score:5, Interesting)

    by XXongo ( 3986865 ) on Friday August 31, 2018 @01:18PM (#57232274) Homepage
    Not to mention that AIs have the opposite problem as well, tagging innocuous posts as hate speach.

    aka the "Scunthorpe problem" [wikipedia.org].

    • Language is difficult, complex, and fluid. We can make meaning out of fairly new and novel combinations of letters and cymbals, and we can parse incorrect words and make meanies out of them. We can also use words in novel ways to make new meaning, and I do notsee AI being able to pick up something like that anytime in the near future.

      Humans have a brain that's wired for understanding language. It's going to be a long time before AI can learn on the vast amount of human skill in this area to be able to do it

  • 1st Amendment (Score:5, Insightful)

    by sdinfoserv ( 1793266 ) on Friday August 31, 2018 @01:23PM (#57232326)
    Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press; or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances." So long as the Constitution Stands, the notion of hate speech is nonsense. to quote an American President: "You want free speech? Let’s see you acknowledge a man whose words make your blood boil, who’s standing center stage and advocating at the top of his lungs that which you would spend a lifetime opposing at the top of yours.” That's freedom. The moment you silence someone, labeling their words or ideas as "hate" is the moment every single person whose died for the Constitution, has died a meaningless death - ending in the death of the promise and the idea of Freedom itself.

    • Re: (Score:2, Insightful)

      by Anonymous Coward

      Congress shall make no law....

      Congress

      Private site, private terms of service. Don't like it, build your own site.

      Some of us don't want to wade through the poisonous angry vitriol spewed by many on just about every site nowadays with a message forum feature. I am glad to see sites do something about it, and I encourage more of it. Automating it is required due to the sheer volume of bile they are confronted with.

      • AOL chat rooms are to the left Perfect for SJWs and other children.

        The internet was built for grownups. It's just not child safe. Pease nerf it with a whitelist of sites with bad words filtered out, and leave the rest of it alone. Maybe the scientologists can 'help' you?

        Also: Protip /.s idiotic 'bad word filter' can be defeated with tags in the words. e.g. Bold, unbold in the middle of words usually filtered.

      • Private site, private terms of service. Don't like it, build your own site.

        Ignoring how "build your own platform (that costs millions of dollars and is wholly unreasonable to expect a private citizen to engage in)" is a bullshit argument, would you also say that about a private business that's "open to the public?"

        E.G. if a baker wants to refuse to bake goods for certain minority demographics, you think they should be able to? Or if a toll road owner wants to bar people of a certain political ideology from travelling the roads they own, they have that right? Just trying to see how

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      Notice how it says nothing about you having to let anyone and everyone into your private establishment and allow them to say whatever they like.

      Get back to us when Facebook.gov is a thing.

      • In a time when newspapers where nailed to the center of town, yes that's true. Today, I would argue, that Facebook has replaced that center of town sounding board. To the degree where it has flourished at the direct demise of old school news papers. News organizations once flush with advertising cash had editors, researchers, fact checkers, etc... and they were/are required by law not to give disproportionate attention to 1 or another candidate.
        Given all the resources, and 10s of thousands of jobs h

    • Re: (Score:2)

      by ljw1004 ( 764174 )

      So long as the Constitution Stands, the notion of hate speech is nonsense. To quote an American President: ...

      That's a blisteringly US-centric response. Here was a paper authored by some Fins and an Italian. From the introduction of the paper, hate speech is defined in both US law and EU law. The paper looked at datasets from Wikipedia and Twitter, both places with a hugely international corpus.

      The paper in any case is independent of what you or the law defines as hate speech. In the paper, they judged whether a variety of published algorithms could come up with the same "hatefulness scoring" as a crowd-sourced col

  • Stop trying to "flag" hate speech (Score:5, Insightful)

    by mysidia ( 191772 ) on Friday August 31, 2018 @01:32PM (#57232420)

    Human language is evolutionary, and humans have a need to express their feelings with others, which can occur through various outlets, but mostly by talking about them.

    If someone feels what you call hatred, and you just don't like that, and you feel they shouldn't be able to express that, so you
    try banning a word or phrase, then what do you think happens?

    Either (A) They find a different mode of expression, and for "Hate speech" that may be bad, since their channel may be taking negative actions in the real world to express their feelings instead of talking about it in a more passive setting. You need to allow so called speakers of "Hate Speech" to be able to express their views in order to be able to successfully have a conversation with those people and possibly reason them to a different position, or at least understand the motivating factors.

    Or (B) People find a different word or phrase or image or euphamism to express the same thing.
    Because language is evolutionary --- existing words will be co-opted, or new words or phrases will be created to express what they wish to express.
    These people who would write "hate speech"; will simply use different words or phrases to express their exact feelings, whatever they can find which
    will avoid flagging the detection system -- because language is evolutionary, in time others will begin to get feelings like the new phrases or words they co-opted
      may now qualify as hate speech, and thus the algorithms fall out of date.

    The real fix is not to try and "block" offensive words or "hate speech", BUT instead to modify the social conventions around the language,
    so that there is no such thing as a verboten word, phrase, or sentence.

  • Here's my hate speech (Score:5, Insightful)

    by blindseer ( 891256 ) <blindseer@@@earthlink...net> on Friday August 31, 2018 @01:39PM (#57232488)

    I hate Nazis. I really really hate Nazis. People willing to round up innocent men, women, and children, and kill them are demonic.

    I hate the KKK. I grew up in a predominately Catholic community and I'd hear stories from my dad and grandparents about the KKK holding rallies. They were outnumbered, and they knew this, so they'd play nice without the masks. They'd be a lot of bark with no bite but in other areas of the USA they'd kill Catholics. Not many people are aware of this but the KKK likely strung up as many Catholics in trees as they did blacks. So, I really hate the KKK. I remember David Duke doing some national tour and ending up in the area to do some speech. A lot of people showed up to hear the idiot talk. I guess he thought he'd get some support in a place that was 99% white, but failed the most basic of demographic testing and seems to have not realized that the crowd was 80%+ Catholic.

    I hate these people and they deserve to burn in hell. What I hate more is restrictions on one's ability to express themselves as they wish. Should David Duke have come around here to speak? Not really, but that's just a failure to recognize his audience. He spoke and he had every right to speak. He got on the local evening news and I got to see all the stunned faces at what he was saying. The guy is an idiot and I'm not going to stop him from exposing his idiocy.

    You "Anti-Fa" people out there need to learn a bit from the quiet resistance that David Duke met 25 or so years ago. As I recall no one raised a sign. Certainly no one raised a fist. Everyone I saw listened to the nonsense and then ignored the bastard. That's what I see a lot on the left/right political spectrum, the left want to shut people up and the right wants them to keep talking. If "hate speech" is such a terrible idea then why fear it being spoken?

    So, there's my hate speech rant. I hate the Nazis, I hate the KKK, and I hate people that want to stop "hate speech". This isn't about "hate speech", this is about stopping the political competition from speaking and that should never be tolerated.

  • I had my letter to The Atlantic tagged by AKSMITH as spam ???

  • "What did he say?"

    "He said the sheriff is nearer."

  • "Hate speech" is just a construct to enable censorship of anything the powers controllig the media, dislike.

    • This is a common nonsense belief, it has a rather specific definition:

      https://en.wikipedia.org/wiki/... [wikipedia.org]

      What good reason could there be for attacking a person or group of people based on their immutable characteristics?

      • What constitutes an attack? If I make the statement "Gender is inextricably linked with an individual's genetics", I would be called out for hate speech in some places, on the grounds that I have "attacked" the gender identity of some self-identified groups. These groups may even in fact have had their existence codified into law at this point. Yet my point is merely a statement of opinion about biology. The label of hate speech is regularly used to shut down opinions that someone dislikes. Some forms

  • Of course it's bad... (Score:4, Insightful)

    by argStyopa ( 232550 ) on Friday August 31, 2018 @03:22PM (#57233230) Journal

    ....can YOU give me a concrete definition of hate speech that is more objective than "language that makes me or someone I like very sad"?

    Is Alex Jones bellowing "Sandy Hook was a fraud" hate speech?
    Is Maxine Waters telling people to aggressively confront Trump supporters hate speech?
    Is a racist redneck saying ""Niggas shouldn't throw stones if you live in a glass house...you should watch your mouth 'cause I'll break your face" hate speech?
    Would it be ok if those words were 50-cent lyrics? (They are.)

    • ....can YOU give me a concrete definition of hate speech that is more objective than "language that makes me or someone I like very sad"?

      Yes, of course. It's not hard unless you're a total moron or being deliberately disingenuous.

      It's dead simple: hate speech is any speech that incites or advocates for physical violence against an identifiable group of people.

      Like I could say "you're an idiot", and that wouldn't be hate speech, even though it makes you sad.

      However, if I said "let's kill all the idiots" it would definitely be hate speech against you.

      • Like I could say "you're an idiot", and that wouldn't be hate speech, even though it makes you sad.

        However, if I said "let's kill all the idiots" it would definitely be hate speech against you.

        At the point where you make an actionable threat against another person or group of people, your speech stops being "speech" and becomes "assault." We don't need any new laws or algorithms to deal with this, because assault is already illegal and well-defined.

        So let's be honest with ourselves - assaultive language is not what most people are calling 'hate speech' these days, and the idea of censoring people online has less to do with actionable threats and more to do with unpopular ideologies.

    • Interesting that you chose Alex Jones and Sandy Hook.

      If denying the Holocaust is hate speech, then the Sandy Hook Massacre must be too. Either that, or we would need to figure out a number of dead people that distinguishes one mass murder from another...especially since some of shooters clearly choose their victims based on their skin colour, sex or religion.

      Personally, I think the best response to hate speech is more speech.

  • Thing humans can't do reliably also can't be done by human designed technology news at 11.

  • The reason it doesn't work is because there are no written rules for what constitutes so-called hate speech. That's by design because that's not what it's about at all. It's about power, pure and simple. People pushing the idea of hate speech insist that they be the ones to define it in a "I'll know it when I see it" fashion. Have you noticed that the very people calling out others for using racist dog-whistles, code, and language seem to be the only ones able to hear the whistle, decode the code, and d

  • You can define hate speech in two ways:
    • If the listener finds a statement offensive.
    • or if the speaker intended to offend with a statement.

    Both are subjective, and in many cases highly dependent on context. If you created an AI which could detect it, then the AI by definition would no longer objective, which would defeat the purpose of trying to invent an AI which can detect it.

    At best, you can create an AI which can flag speech would might be considered hate speech. Then leave it up to people (idea

  • Unfortunately, the extreme left, the same extreme left that is pushing 'sane lefties' to the center, has really tried to redefine hate speech in the past 3 to 5 years online.

    If you were to say something genuinely horrible like:

    "I think all homosexuals are mentally deranged and should be exterminated, I emplore people to go out and murder one today!"
    I'd be pretty comfortable saying, that's some pretty shitty, unkind and genuine hate speech.

    However, the modern day definition could be something like:

    "I disagre

Slashdot Top Deals

New York... when civilization falls apart, remember, we were way ahead of you. - David Letterman

Close