9 ms·

Elon “free speech” Musk. >Negative/hate tweets will be max deboosted & demonetized. Guess this guy finally realized that a free for all model is not feasible.

elgenioso 4y ago
It's impossible to detect negative sentiment purely by text. It's an unsolvable problem.
- zamalek 4y ago
  It's more fundamental than that. Some forms of speech can indirectly or directly limit the speech of others. Free speech absolutism in an unachievable panacea; it is a logical fallacy.
  - hellfish 4y ago
    The people who oppose free speech always try to redefine it this way. Freedom of speech is freedom from censorship or moderation. Not freedom from discomfort
    - pixl97 4y ago
      Like the discomfort from being banned from someones private property from saying dumb crap?
      - at-w 4y ago
        Like Comcast deciding they don't want workers discussing unionizing using the internet infrastructure that they privately own. Or financial institutions like PayPal freezing and seizing your money on their "private property" because of speech they don't like. Or when AWS, Microsoft, and the largest and most powerful corporations in human history coordinated to take a separate social network offline because of speech they didn't like. Or when Twitter and Facebook coordinated to falsely label a true story "Russian disinformation" days before an election, literally preventing people from sharing it with each other privately on the world's largest communications networks. Whether Hunter Biden smoking crack with prostitutes would impact the election, or is even anyone's business, is questionable to me. But so is having billionaire tech CEOs decide what people can discuss through private communications with the express intent of suppressing information related to a political campaign. The same goes for credit card monopolies like VISA and Mastercard coordinating to destroy businesses like Mindgeek because of moralist outrage from another "side."
  - ibejoeb 4y ago
    I'm not following you. What can I say that prevents you from saying something? That would seem to violate causality.
    - zamalek 4y ago
      That causal violation is the exact problem with free speech aboslutism. "You are not allowed to criticize the CCP" is speech. "I order you to kill/hang that vice president" is speech, and the dead don't speak.
      - mh- 4y ago
        That last one would already fail the Brandenburg test [0], wouldn't it? (I'm not a lawyer.) [0] https://www.criminaldefenselawyer.com/resources/inciting-to-riot-violence-or-insurrection.html https://www.criminaldefenselawyer.com/resources/inciting-to-...
        zamalek 4y ago
        > Brandenburg test The Brandenburg Test doesn't apply to free speech absolutism.
        erichocean 4y ago
        Elon Musk has already said he supports unrestricted *legal* speech on Twitter, and that individual cultures should decide what that is for themselves, through their local legal system. He's not a "free speech absolutist."
        krapp 4y ago
        He calls himself a free speech absolutist, and he's already banned parody, racist but legal (in the US) speech and criticism of himself and his technical knowledge - which is more restrictive than Twitter was prior. As is the plan to "deboost" negative and hateful tweets - I'll remind the court that when Youtube did the same thing, many people here considered even demonetizing and downranking content to be censorship, and antithetical to free speech.
  - elgenioso 4y ago
    I don't think that's true. The only way to limit speech on an algorithmicially mediated platform is via algorithms and there is no algorithm that can encode non-trivial properties of language regardless of what the AI maximalists might say to the contrary.
- lrvick 4y ago
  As someone who specialized in this area for years in social media specifically, I have never seen evidence you can get past 80% accuracy with any known sentiment analysis method. Brute force stats like Naive Bayes is still as good as it gets on platforms like Twitter where speech patterns are too short and irregular for traditional NLP to be very useful.
  - elgenioso 4y ago
    Even non-traditional NLP won't work. There is no way to algorithmically enforce "good" speech. I agree that the character limitations don't help and the 280 character limitation is the main reason Twitter is such a mess. There are no coherent long form texts generated by AI models but 280 characters is very easy because people will give the author the benefit of the doubt and interpolate the missing context.
  - MichaelCollins 4y ago
    > 80% accuracy Sentiment analysis is way worse than that in adversarial scenarios, where a human is trying to outwit the machine. My prediction for this whole thing: Elon Musk [again] gets swept up by AI hype, tries to replace humans with software, and it doesn't work. Like at Tesla, so at Twitter.
- garyfirestorm 4y ago
  True! How would it detect sarcasm. For eg. I love what he is doing to Twitter
  - lrvick 4y ago
    In my experience people most often convey sarcasm on social media with emoji and the emoji is typically the most informative feature. I hate my job :) -> probably loves their job. I love my job :( -> probably hates their job. Without emoji sarcasm is often hard to detect even for a human.
    - throwaway290 4y ago
      It's exceedingly easy to detect sentiment for any human who cares and follows context, next to impossible to do automatically. "Good job Elon :)" can mean diametrically opposing things depending on author's views and what they wrote up to this moment. Same with "I love my job :)", etc. Employing an AI solution could cause some fun blunders but would likely amplify the amount of burning sarcasm overall.
    - rcoveson 4y ago
      This is wrong as often as it's right. I predect that "I hate my job :)" on social media really does mean the speaker hates their job at least 40% of the time. It's sort of a "this is fine"[0] smiley; the smiley is the sarcasm while the statement is literally true. 0. https://knowyourmeme.com/memes/this-is-fine https://knowyourmeme.com/memes/this-is-fine
    - coffeebeqn 4y ago
      Elon is doing a great job at Twitter :) What do I really mean?
- paxys 4y ago
  Having Elon in sole control of what is or isn’t acceptable speech is a feature, not a bug.
- bee_rider 4y ago
  Presumably he’s going to realize this in a couple days and start trying to hire back the folks who were in charge of moderation, etc, on the platform. The possibility that Twitter was configured the way it was when he bought it for a reason apparently didn’t occur to him.
- coffeebeqn 4y ago
  I mean that whole idea sounds extremely dystopian. You can only post things that the AI considers “positive”. Putin is a piece of shit who bombed my family! I’m sorry Dave, I can’t allow you to post that.
GloriousKoji 4y ago
Free as in capital, not beer.
mistermann 4y ago
> Elon “free speech” Musk. Is your usage of quotations marks intended to adhere to one of these? If so, which one, and how? If not, what is the intent of them? ------------------------------------------------- https://www.grammarbook.com/punctuation/quotes.asp https://www.grammarbook.com/punctuation/quotes.asp - Rule 1. Use double quotation marks to set off a direct (word-for-word) quotation. Correct: “I hope you will be here,” he said. Incorrect: He said that he “hoped I would be there.” (The quotation marks are incorrect because hoped I would be there does not state the speaker's exact words.) - Rule 8a. Quotation marks are often used with technical terms, terms used in an unusual way, or other expressions that vary from standard usage. Examples: It's an oil-extraction method known as “fracking.” He did some “experimenting” in his college days. I had a visit from my “friend” the tax man.
- throwaway290 4y ago
  If you are serious, it's a common way of specifying a nickname.
  - mistermann 4y ago
    I am serious, and it is indeed that sometimes...but I wonder if that's the intent Ztynovovk had when making the comment (or: what it "is" from within the initial frame of reference, as "is-ness" is relative to the observer). I think it's fun and informative to draw attention to cognitive functionality of humans when they are opining on the cognitive functionality of other humans, particularly in "smart" communities when the topic of discussion tends to exploit cultural norms and catalyze ~unforced (note: speculation) cognitive errors.
- Ztynovovk 4y ago
  They indicate how much of a clown he is.
- naniwaduni 4y ago
  > terms used in an unusual way > I had a visit from my “friend” the tax man. If you're going to pull out the dictionary in bad faith, it would behoove you to read the defs you're pulling out.
googlryas 4y ago
Is there any reason to naively interpret his view of free speech as "anyone can say anything at any time"?
tootie 4y ago
What the hell does "demonetize" even mean? YouTube uses that term to mean they won't pay the creator. Twitter doesn't pay anyone. Removing ads just makes the content better.

elgenioso 4y ago

It's impossible to detect negative sentiment purely by text. It's an unsolvable problem.

zamalek 4y ago
It's more fundamental than that. Some forms of speech can indirectly or directly limit the speech of others. Free speech absolutism in an unachievable panacea; it is a logical fallacy.
- hellfish 4y ago
  The people who oppose free speech always try to redefine it this way. Freedom of speech is freedom from censorship or moderation. Not freedom from discomfort
  - pixl97 4y ago
    Like the discomfort from being banned from someones private property from saying dumb crap?
    - at-w 4y ago
      Like Comcast deciding they don't want workers discussing unionizing using the internet infrastructure that they privately own. Or financial institutions like PayPal freezing and seizing your money on their "private property" because of speech they don't like. Or when AWS, Microsoft, and the largest and most powerful corporations in human history coordinated to take a separate social network offline because of speech they didn't like. Or when Twitter and Facebook coordinated to falsely label a true story "Russian disinformation" days before an election, literally preventing people from sharing it with each other privately on the world's largest communications networks. Whether Hunter Biden smoking crack with prostitutes would impact the election, or is even anyone's business, is questionable to me. But so is having billionaire tech CEOs decide what people can discuss through private communications with the express intent of suppressing information related to a political campaign. The same goes for credit card monopolies like VISA and Mastercard coordinating to destroy businesses like Mindgeek because of moralist outrage from another "side."
- ibejoeb 4y ago
  I'm not following you. What can I say that prevents you from saying something? That would seem to violate causality.
  - zamalek 4y ago
    That causal violation is the exact problem with free speech aboslutism. "You are not allowed to criticize the CCP" is speech. "I order you to kill/hang that vice president" is speech, and the dead don't speak.
    - mh- 4y ago
      That last one would already fail the Brandenburg test [0], wouldn't it? (I'm not a lawyer.) [0] https://www.criminaldefenselawyer.com/resources/inciting-to-riot-violence-or-insurrection.html https://www.criminaldefenselawyer.com/resources/inciting-to-...
      - zamalek 4y ago
        > Brandenburg test The Brandenburg Test doesn't apply to free speech absolutism.
        erichocean 4y ago
        Elon Musk has already said he supports unrestricted *legal* speech on Twitter, and that individual cultures should decide what that is for themselves, through their local legal system. He's not a "free speech absolutist."
        krapp 4y ago
        He calls himself a free speech absolutist, and he's already banned parody, racist but legal (in the US) speech and criticism of himself and his technical knowledge - which is more restrictive than Twitter was prior. As is the plan to "deboost" negative and hateful tweets - I'll remind the court that when Youtube did the same thing, many people here considered even demonetizing and downranking content to be censorship, and antithetical to free speech.
- elgenioso 4y ago
  I don't think that's true. The only way to limit speech on an algorithmicially mediated platform is via algorithms and there is no algorithm that can encode non-trivial properties of language regardless of what the AI maximalists might say to the contrary.
lrvick 4y ago
As someone who specialized in this area for years in social media specifically, I have never seen evidence you can get past 80% accuracy with any known sentiment analysis method. Brute force stats like Naive Bayes is still as good as it gets on platforms like Twitter where speech patterns are too short and irregular for traditional NLP to be very useful.
- elgenioso 4y ago
  Even non-traditional NLP won't work. There is no way to algorithmically enforce "good" speech. I agree that the character limitations don't help and the 280 character limitation is the main reason Twitter is such a mess. There are no coherent long form texts generated by AI models but 280 characters is very easy because people will give the author the benefit of the doubt and interpolate the missing context.
- MichaelCollins 4y ago
  > 80% accuracy Sentiment analysis is way worse than that in adversarial scenarios, where a human is trying to outwit the machine. My prediction for this whole thing: Elon Musk [again] gets swept up by AI hype, tries to replace humans with software, and it doesn't work. Like at Tesla, so at Twitter.
garyfirestorm 4y ago
True! How would it detect sarcasm. For eg. I love what he is doing to Twitter
- lrvick 4y ago
  In my experience people most often convey sarcasm on social media with emoji and the emoji is typically the most informative feature. I hate my job :) -> probably loves their job. I love my job :( -> probably hates their job. Without emoji sarcasm is often hard to detect even for a human.
  - throwaway290 4y ago
    It's exceedingly easy to detect sentiment for any human who cares and follows context, next to impossible to do automatically. "Good job Elon :)" can mean diametrically opposing things depending on author's views and what they wrote up to this moment. Same with "I love my job :)", etc. Employing an AI solution could cause some fun blunders but would likely amplify the amount of burning sarcasm overall.
  - rcoveson 4y ago
    This is wrong as often as it's right. I predect that "I hate my job :)" on social media really does mean the speaker hates their job at least 40% of the time. It's sort of a "this is fine"[0] smiley; the smiley is the sarcasm while the statement is literally true. 0. https://knowyourmeme.com/memes/this-is-fine https://knowyourmeme.com/memes/this-is-fine
  - coffeebeqn 4y ago
    Elon is doing a great job at Twitter :) What do I really mean?
paxys 4y ago
Having Elon in sole control of what is or isn’t acceptable speech is a feature, not a bug.
bee_rider 4y ago
Presumably he’s going to realize this in a couple days and start trying to hire back the folks who were in charge of moderation, etc, on the platform. The possibility that Twitter was configured the way it was when he bought it for a reason apparently didn’t occur to him.
coffeebeqn 4y ago
I mean that whole idea sounds extremely dystopian. You can only post things that the AI considers “positive”. Putin is a piece of shit who bombed my family! I’m sorry Dave, I can’t allow you to post that.

GloriousKoji 4y ago

Free as in capital, not beer.

mistermann 4y ago

> Elon “free speech” Musk. Is your usage of quotations marks intended to adhere to one of these? If so, which one, and how? If not, what is the intent of them? ------------------------------------------------- https://www.grammarbook.com/punctuation/quotes.asp https://www.grammarbook.com/punctuation/quotes.asp - Rule 1. Use double quotation marks to set off a direct (word-for-word) quotation. Correct: “I hope you will be here,” he said. Incorrect: He said that he “hoped I would be there.” (The quotation marks are incorrect because hoped I would be there does not state the speaker's exact words.) - Rule 8a. Quotation marks are often used with technical terms, terms used in an unusual way, or other expressions that vary from standard usage. Examples: It's an oil-extraction method known as “fracking.” He did some “experimenting” in his college days. I had a visit from my “friend” the tax man.

throwaway290 4y ago
If you are serious, it's a common way of specifying a nickname.
- mistermann 4y ago
  I am serious, and it is indeed that sometimes...but I wonder if that's the intent Ztynovovk had when making the comment (or: what it "is" from within the initial frame of reference, as "is-ness" is relative to the observer). I think it's fun and informative to draw attention to cognitive functionality of humans when they are opining on the cognitive functionality of other humans, particularly in "smart" communities when the topic of discussion tends to exploit cultural norms and catalyze ~unforced (note: speculation) cognitive errors.
Ztynovovk 4y ago
They indicate how much of a clown he is.
naniwaduni 4y ago
> terms used in an unusual way > I had a visit from my “friend” the tax man. If you're going to pull out the dictionary in bad faith, it would behoove you to read the defs you're pulling out.

googlryas 4y ago

Is there any reason to naively interpret his view of free speech as "anyone can say anything at any time"?

tootie 4y ago

What the hell does "demonetize" even mean? YouTube uses that term to mean they won't pay the creator. Twitter doesn't pay anyone. Removing ads just makes the content better.