3 ms·

If the executives did not make a meaningful effort to count them, that is fairly damning, given how much the stock price swings on the count. Nobody said it wa

If the executives did not make a meaningful effort to count them, that is fairly damning, given how much the stock price swings on the count. Nobody said it was easy, but it's certainly harder if you don't try.

zimpenfish 4y ago
> If the executives did not make a meaningful effort to count them They've been filing their methodology for bot counting with the SEC since 2013. If they're not making a "meaningful effort" and it materially affected the stock price in some way, either the SEC or a shareholder would have gone "HOLD ON SHENANIGANS O'CLOCK", surely? It can't be that the entire world was A-OK with Twitter's bot counting until June 2022 when a man claiming to want to buy Twitter to fix the bot problem got cold feet on a market drop...
- frumper 4y ago
  Shenanigans can go on for a lot longer than 9 years without anyone noticing.
- bpodgursky 4y ago
  The "methodology" is that people look at 100 accounts a day and determine whether they are bots. They have never disclosed any of the signals that go into this determination. You have a lot of faith in the immediately efficient market here.
  - Cederfjard 4y ago
    The point is that they have not claimed anything regarding this in their filings that isn't true, not whether or not you think they've been clear and detailed enough to answer the question properly. And to give Musk an out, which is what this tangent is about, not only do they need to have actually lied, the lies need to have had a VERY substantial effect on the price of the company. The bot thing simply does not help Musk get out of the deal he's made. That is not the same thing as "Twitter are great at dealing with bots and have been very transparent about how they do it", but that's not the bar that has to be cleared here.
- happyopossum 4y ago
  > They've been filing their methodology for bot counting with the SEC since 2013. No, they haven’t. They describe at a very high level the amount of sampling they do (100 accounts a day? Really, that’s it?), but don’t discuss the methodology used, such as what they use as signals and indicators of botness. That’s not “filing their methodology“, it’s covering their arses.
  - Sebb767 4y ago
    > That’s not “filing their methodology“, it’s covering their arses. True, but probably quite successful. So this will most likely not save Musk.
  - jcranmer 4y ago
    > (100 accounts a day? Really, that’s it?) Today's question on your statistics 101 exam: You have a population of 100 million people. You estimate that the true probability of some statistic is about 5%. What sample size do you need to be 95% sure that you are within 5% of the correct answer? Answer: 73. (No really, this kind of question is absolutely going to be in a Stats 101 class. And sample sizes really don't need to be that big to be accurate.)
- the_doctah 4y ago
  How is it any harder than giving users a captcha?
  - Sebb767 4y ago
    Captcha solvers exist and are quite accessible. If solving the problem of bots would be as easy as showing a captcha, we would not have bot problems.

zimpenfish 4y ago

> If the executives did not make a meaningful effort to count them They've been filing their methodology for bot counting with the SEC since 2013. If they're not making a "meaningful effort" and it materially affected the stock price in some way, either the SEC or a shareholder would have gone "HOLD ON SHENANIGANS O'CLOCK", surely? It can't be that the entire world was A-OK with Twitter's bot counting until June 2022 when a man claiming to want to buy Twitter to fix the bot problem got cold feet on a market drop...

frumper 4y ago
Shenanigans can go on for a lot longer than 9 years without anyone noticing.
bpodgursky 4y ago
The "methodology" is that people look at 100 accounts a day and determine whether they are bots. They have never disclosed any of the signals that go into this determination. You have a lot of faith in the immediately efficient market here.
- Cederfjard 4y ago
  The point is that they have not claimed anything regarding this in their filings that isn't true, not whether or not you think they've been clear and detailed enough to answer the question properly. And to give Musk an out, which is what this tangent is about, not only do they need to have actually lied, the lies need to have had a VERY substantial effect on the price of the company. The bot thing simply does not help Musk get out of the deal he's made. That is not the same thing as "Twitter are great at dealing with bots and have been very transparent about how they do it", but that's not the bar that has to be cleared here.
happyopossum 4y ago
> They've been filing their methodology for bot counting with the SEC since 2013. No, they haven’t. They describe at a very high level the amount of sampling they do (100 accounts a day? Really, that’s it?), but don’t discuss the methodology used, such as what they use as signals and indicators of botness. That’s not “filing their methodology“, it’s covering their arses.
- Sebb767 4y ago
  > That’s not “filing their methodology“, it’s covering their arses. True, but probably quite successful. So this will most likely not save Musk.
- jcranmer 4y ago
  > (100 accounts a day? Really, that’s it?) Today's question on your statistics 101 exam: You have a population of 100 million people. You estimate that the true probability of some statistic is about 5%. What sample size do you need to be 95% sure that you are within 5% of the correct answer? Answer: 73. (No really, this kind of question is absolutely going to be in a Stats 101 class. And sample sizes really don't need to be that big to be accurate.)
the_doctah 4y ago
How is it any harder than giving users a captcha?
- Sebb767 4y ago
  Captcha solvers exist and are quite accessible. If solving the problem of bots would be as easy as showing a captcha, we would not have bot problems.