in my experience most tech ceos just use platitudes constantly. having worked at a few large companies, they do it internally too. that way they have latitude to do anything they want
From the Machiavellian / realpolitik view, there is just little incentive for them to be fully transparent.
Partly, because we can't hold them to account anyhow, but also because many of us cannot handle the truth, i.e. 'we don't want to see how the sausage is made.'
Twitter now has the utterly impossible task of minding/nannying the language used by millions of people, in social and legal context. Whenever they get it 'wrong' (by some journalists standard) they get blasted.
It's possible they could be doing it better, but it's nary an impossible challenge, at least not without some fundamental changes to the service.
Do we really want Jack and Zuck having the final say on what we can and cannot do?
I'm sure they do not want this responsibility. They've just painted themselves into a corner with it.
Given how deeply contextualized information is ... I'm even doubtful AI is on the horizon. And may of the 'decisions' made to ban posts/accounts theoretically require some heavy and contemplate deliberation, and even legal analysis.
It just doesn't work, we need to re-think all of this.
> I'm sure they do not want this responsibility. They've just painted themselves into a corner with it.
Indeed, they needed to be more honest upfront and say that the demands from certain quarters aren't possible. That's a hard thing to be honest about, but at least you'd be setting realistic expectations.
The 'correct answer' is "We're not doing so well, but we're working on it". Just pretend to be humble and waive your hands until the press moves onto some other narrative.
Agreed. I dont think this paper was the one I was looking for, but it will do. [1] The authors are working on handling negation in sentiment analysis. Ie, this movie is good, this movie is not good, this movie is not very good.
"One of the most important sub tasks in sentiment analysis is to
determine the sequence of words affected by negation. Most of the existing sentiment analysis systems used
traditional methods based on static window and punctuation marks to determine the scope of negation.
However, these methods do not offer satisfactory performance due to variability in the negation scope,
inability to deal with linguistic features and improper word sense disambiguation. " [2]
[1] Sorry to be in a hurry, i just think this point is important given the hype around AI, and all the talk about using AI to "manage" human conversations. Look up AI winter if you want to learn about hype, AI, and expectations :)
[2] http://www.jcomputers.us/vol12/jcp1205-11.pdf
For example, in order for sentiment to be 'racist' - the ethnic nature of the participants may have to be known.
So it's not about 'analyzing a statement' so much as it is understanding so much about people, individuals, current thresholds for tolerance, current events. Etc. etc..
It's impossible to look at a 'post' in total isolation, it has to be put in the context of the real world which is way beyond AI today.
So it's not possible to 'train' AI simply by putting 'sentiment' as an input parameter. You'd have to 'train' NN's with inputs that include the entirety of CNN, Wikipedia and pop culture!
He's purposefully vague, because there's little to gain from taking a stance. The goal of these recent interviews is to make him more relatable, it's a fluffy PR tour.
Can we punctuate this title so that it might not mean that Swisher is interviewing a transcript? Putting the last word in parentheses would achieve that.
> I want people to walk away from Twitter feeling like they learned something and feeling empowered to some degree. It depresses me when that’s not the general vibe, and inspires me to figure it out. That’s my desire
Last year, in March I deleted all my tweets and closed my account [1]. A month ago I discovered that my account appears as "suspended" [2]. I don't feel empowered by this Tweeter information about me.
Perhaps, but when someone tries to open an account then Twitter checks first if the name is taken. Therefore how hard would it be for them to keep all the used names and verify first?
This is just a hypothesis, only Twitter knows. Another hypothesis is that Twitter wants to present those who leave as suspended.
> Once you’ve changed your username, it returns to general circulation and may be selected by another account.
So in the case of services where your username is the only thing your account is identified on, it probably makes more sense to delete your tweets and leave the account active. Oh, and post a nonsense message every few months so you don't get deactivated
You are right. Moreover, never give any personal information to any social media corporation, right? There is also an advantage for Twitter: they can claim that they still have active users which are not bots.
It's very easy to keep a list of all used names and verify first. Twitter does that. You requested to have your account removed from that list. What would even be the point of keeping your username locked after the account deletion?
What's the point? Then why does Twitter verify if there is another account by the same name name when one opens an account? They could just associate the password with a concatenation of the username with a random number. The same point.
Er, but the account name you used to use wasn't taken, because you deleted your account. Doesn't twitter make it clear in their documentation that if you delete your account, your account name is returned to the pool of available account names? And, if that is the case, I don't think it's an unreasonable way of handling the situation.
Re: your hypothesis about twitter wanting to present everyone who leaves as suspended: This is highly, highly unlikely and you'd have to provide significant evidence to the contrary.
All evidence is at Twitter. There is no way to verify any hypothesis without access to what Twitter has, including the hypothesis that the most users on Twitter are not bots. But everybody is free to have any opinion based on the interaction they have with Twitter. In my case, for a reason which is completely under the control of Twitter, my username which can be linked with my real identity is presented as suspended, when in fact I deleted my account. I don't longer have any relation with Twitter but they present an account which contains my name as suspended. Legally true, morally repugnant.
Say you ran a website once. Then you let your domain expire. Someone else bought it, and put up their site on the domain. Would you blame the domain registrar for that?
More like you live in another country and you leave and later discover that when people who know you there look for you, they find out your are on a list of offenders, for a reason known only by the police from that country.
To close this discussion, I propose a law (maybe in EU) that any interaction with a social media corporation has to be mediated by a provable legal advice from a professional. In this way all these discussions which mix the moral side (users) and the legal side (corporation) will be avoided :)
“Kara: I am going to start a NEW thread to make it easy for people to follow (@waltmossberg just texted me that it is a "chaotic hellpit"). Stay in that one. OK?”
Being able to have a public chat that others could follow was a solved problem back in 1994. How the hell is this dystopian nightmare of a conversation happening in the year 2019?
I see some of the merits of Twitter, but in general, I believe Twitter's net effect is a dramatic deevolution of conversation.
We now have popular third-party services, such as Thread Reader App, just to make content readable again posted by people who are writing prose through SMS.
Strident would be good if there was anything behind it, but she never says "this is over until you answer the question" or anything to that effect, so Dorsey just waffles past it.
If I were a tech exec in hot water for selling out civilization for billions in net losses, I'd be begging her to interview me. Nobody can make you feel sympathy for the devil like Kara can.
> Jack: I also don’t feel good about how Twitter tends to incentivize outrage, fast takes, short term thinking, echo chambers, and fragmented conversation and consideration. Are they fixable? I believe we can do a lot to address. And likely have to change more fundamentals to do so.
While my initial reaction was to be positively surprised that he acknowledges these problems, I can't help but be cynical and read this as if Big Tobacco were to argue that they need to get rid of nicotine.
outrage, fast takes, short term thinking, echo chambers... that's what's getting people hooked on Twitter.
Well the weird thing is that Twitter is a network defined by length limits on messages. It exists because it forbids users from raising the bar by writing anything long enough to be interesting, intelligent, insightful or polite. Now he's surprised at the outcome? It was predictable from the very start!
That’s really messy. How do people follow a discussion like this one? You basically have to follow a tree of threads but you only see one chunk at a time, this have to keep in memory the full context of previous nodes when you jump back. That’s truly hard to follow. In HN for example the level of indentation shows you the current depth, so you always have a kind of global view of the full tree.
> What do you think about twitter breaks and purges. Why do you think that is? I can’t say I’ve heard many people say they feel “good” after not being on twitter for a while
I've heard nothing but good things from people who expunge all social media from their lives, myself included. Social media is the problem, not whatever she thinks he can possibly do (other than shutter their doors).
You are posting in a public forum on Twitter, and anyone can speak or threaten you and twenty other people internationally while simultaneously playing a board game, watching TV, and screaming at their kids. You expect Twitter to be able to control in even the slightest the communications of potentially 7 billion people?
I think the problem is that Jack Dorsey decided he has the same responsibilities as Mark Zuckerberg. Facebook is an entirely different advertisement ridden animal that Mark created features to essentially help bad actors do their thing. Twitter is a communications platform much like standard texting is (just public, which adds more complexity to the gumbo), yet AT&T and Verizon have never needed to claim responsibility for people getting threatened via texting.
Kara: Why not just be more stringent on kicking off offenders? It seems like you tolerate a lot. If Twitter ran my house, my kids would be eating ramen, playing Red Dead Redemption 2 and wearing filthy socks
I feel like Kara Swisher is the most misunderstood tech reporter. In the sense that outsiders think she makes a lot of sense. SHE DOESN'T. This question right here shows her ignorance of tech (how insanely challenging it is to track every damn thing in the whole world and make out what's/who's right/wrong).
Kicking off offenders? Let her please enlighten us on how to decide on who to kick off. And she will be the first one with a placard "FREEDOM OF SPEECH" if someone of her fancy gets blocked on Twitter or similar sites for saying things that SHE likes.
The hypocrisy of these people pleasing reporters is just mind boggling. Specially for someone at her level. (she had got Steve Jobs and Bill Gates together on stage).
The speed of making progress was discussed. I feel like twitter has slowed down after they switched away from Rails. Maybe technically speaking they are having problem with Scala in terms of development speed? Does anyone know the specifics?
57 comments
[ 2.6 ms ] story [ 114 ms ] threadPartly, because we can't hold them to account anyhow, but also because many of us cannot handle the truth, i.e. 'we don't want to see how the sausage is made.'
Twitter now has the utterly impossible task of minding/nannying the language used by millions of people, in social and legal context. Whenever they get it 'wrong' (by some journalists standard) they get blasted.
It's possible they could be doing it better, but it's nary an impossible challenge, at least not without some fundamental changes to the service.
Do we really want Jack and Zuck having the final say on what we can and cannot do?
I'm sure they do not want this responsibility. They've just painted themselves into a corner with it.
Given how deeply contextualized information is ... I'm even doubtful AI is on the horizon. And may of the 'decisions' made to ban posts/accounts theoretically require some heavy and contemplate deliberation, and even legal analysis.
It just doesn't work, we need to re-think all of this.
Indeed, they needed to be more honest upfront and say that the demands from certain quarters aren't possible. That's a hard thing to be honest about, but at least you'd be setting realistic expectations.
The 'correct answer' is "We're not doing so well, but we're working on it". Just pretend to be humble and waive your hands until the press moves onto some other narrative.
Agreed. I dont think this paper was the one I was looking for, but it will do. [1] The authors are working on handling negation in sentiment analysis. Ie, this movie is good, this movie is not good, this movie is not very good.
"One of the most important sub tasks in sentiment analysis is to determine the sequence of words affected by negation. Most of the existing sentiment analysis systems used traditional methods based on static window and punctuation marks to determine the scope of negation. However, these methods do not offer satisfactory performance due to variability in the negation scope, inability to deal with linguistic features and improper word sense disambiguation. " [2]
[1] Sorry to be in a hurry, i just think this point is important given the hype around AI, and all the talk about using AI to "manage" human conversations. Look up AI winter if you want to learn about hype, AI, and expectations :) [2] http://www.jcomputers.us/vol12/jcp1205-11.pdf
But the more important issue will be context.
For example, in order for sentiment to be 'racist' - the ethnic nature of the participants may have to be known.
So it's not about 'analyzing a statement' so much as it is understanding so much about people, individuals, current thresholds for tolerance, current events. Etc. etc..
It's impossible to look at a 'post' in total isolation, it has to be put in the context of the real world which is way beyond AI today.
So it's not possible to 'train' AI simply by putting 'sentiment' as an input parameter. You'd have to 'train' NN's with inputs that include the entirety of CNN, Wikipedia and pop culture!
What a mess.
Funny how she bitched about zuck even though he does pretty well these days in interviews (hell even when he had to endure the congress hearing)
https://twitter.com/jack/status/1071575088695140353
(NSFW - language) https://www.youtube.com/watch?v=h5d82ndui_s
Last year, in March I deleted all my tweets and closed my account [1]. A month ago I discovered that my account appears as "suspended" [2]. I don't feel empowered by this Tweeter information about me.
[1] https://chorasimilarity.wordpress.com/2018/03/21/i-deleted-f...
[2] https://chorasimilarity.wordpress.com/2019/01/01/twitter-lie...
This is just a hypothesis, only Twitter knows. Another hypothesis is that Twitter wants to present those who leave as suspended.
> Once you’ve changed your username, it returns to general circulation and may be selected by another account.
So in the case of services where your username is the only thing your account is identified on, it probably makes more sense to delete your tweets and leave the account active. Oh, and post a nonsense message every few months so you don't get deactivated
Re: your hypothesis about twitter wanting to present everyone who leaves as suspended: This is highly, highly unlikely and you'd have to provide significant evidence to the contrary.
https://twitter.com/Flexi23/status/973219244233158662
This kind of misrepresentation may be legal (or not, IANAL) but I find it infuriating.
“Kara: I am going to start a NEW thread to make it easy for people to follow (@waltmossberg just texted me that it is a "chaotic hellpit"). Stay in that one. OK?”
Being able to have a public chat that others could follow was a solved problem back in 1994. How the hell is this dystopian nightmare of a conversation happening in the year 2019?
We now have popular third-party services, such as Thread Reader App, just to make content readable again posted by people who are writing prose through SMS.
If anyone could prevent Jack Dorsey from speaking in generalities, I'd assume it would be she.
If I were a tech exec in hot water for selling out civilization for billions in net losses, I'd be begging her to interview me. Nobody can make you feel sympathy for the devil like Kara can.
- Kara Swisher, in the sub-headline of a recent gossip column published in the NYT opinion section on Jeff Bezos’s dicorce drama
Technological illiteracy and fear-mongering should be opposed at every turn.
https://samharris.org/podcasts/148-jack-dorsey/
While my initial reaction was to be positively surprised that he acknowledges these problems, I can't help but be cynical and read this as if Big Tobacco were to argue that they need to get rid of nicotine.
outrage, fast takes, short term thinking, echo chambers... that's what's getting people hooked on Twitter.
I've heard nothing but good things from people who expunge all social media from their lives, myself included. Social media is the problem, not whatever she thinks he can possibly do (other than shutter their doors).
You are posting in a public forum on Twitter, and anyone can speak or threaten you and twenty other people internationally while simultaneously playing a board game, watching TV, and screaming at their kids. You expect Twitter to be able to control in even the slightest the communications of potentially 7 billion people?
I think the problem is that Jack Dorsey decided he has the same responsibilities as Mark Zuckerberg. Facebook is an entirely different advertisement ridden animal that Mark created features to essentially help bad actors do their thing. Twitter is a communications platform much like standard texting is (just public, which adds more complexity to the gumbo), yet AT&T and Verizon have never needed to claim responsibility for people getting threatened via texting.
I feel like Kara Swisher is the most misunderstood tech reporter. In the sense that outsiders think she makes a lot of sense. SHE DOESN'T. This question right here shows her ignorance of tech (how insanely challenging it is to track every damn thing in the whole world and make out what's/who's right/wrong).
Kicking off offenders? Let her please enlighten us on how to decide on who to kick off. And she will be the first one with a placard "FREEDOM OF SPEECH" if someone of her fancy gets blocked on Twitter or similar sites for saying things that SHE likes.
The hypocrisy of these people pleasing reporters is just mind boggling. Specially for someone at her level. (she had got Steve Jobs and Bill Gates together on stage).
Also ramen and RDR2? Golden combo. You go kids!