5 ms·

Have you seen credible evidence that they are right about their numbers? Without Twitter providing more context and showing their work, the number can't be tru

Have you seen credible evidence that they are right about their numbers? Without Twitter providing more context and showing their work, the number can't be trusted, thus it isn't credible. If you claim you have such a low number of bots (5%), you better prove it, otherwise I'll assume the worst and go with 20%. Anyone that goes on Twitter knows it's above 5%. All 3rd party estimations are well above 5%. Prove it Twitter. Prove it.

pcmoney 4y ago
They literally never claimed this was accurate and even said their methodology could be wrong. Nothing in the contract says anything about bots. They don’t have to prove anything at all.
- hunterb123 4y ago
  > They literally never claimed this was accurate > and even said their methodology could be wrong Yeah that's my point, their supplied number is unusable for investors, advertisers, etc. Combine that with the bot experiences on Twitter, the 3rd party estimates of 10-20% if looks way off. I don't see why NOW anyone would trust 5%. Diligence has NOW been done and that "methodology could be wrong" looks likely. -- EDIT (post limit) -- > But that's the whole point - Musk waived his right to do due diligence, so he is pointing at this statement being wrong (and not unusable) as his way out. @suresk, that's not a thing, but it is a talking point, but I'm not debating that. I'm saying NOW that the diligence is done and we know it's probably not accurate. See my first post.
  - suresk 4y ago
    But that's the whole point - Musk waived his right to do due diligence, so he is pointing at this statement being wrong (and not unusable) as his way out.
  - Analemma_ 4y ago
    People keep trying to explain to you why you're wrong and you keep on ignoring them and persisting with your original false statement. The 5% number claimed by Twitter is not the percentage of all accounts that are bots. That has never been the claim. Twitter separates its accounts into "mDAU"s and "other", and they are claiming 5% of the mDAUs are bots. You cannot falsify this claim with "bot experiences on Twitter", because all the bots you're seeing could be correctly classified as "other". Indeed, nobody outside Twitter can falsify this claim because it's a statement about Twitter's internal classification. The only thing that can falsify Twitter's claim is their own internal data, so that's why discovery is being done.
    - hunterb123 4y ago
      I'm not talking about all bots. I'm talking about mDAU bots. You guys are trying to argue against a point I'm not making. 3rd party estimations range from 10-20% for mDAU bots, Twitter's is 5%. Others were defending it by saying "well the number doesn't have to be accurate legally" or "well Musk should have known it's not really accurate". My counter is that weasel legal wording and logic doesn't give much confidence in Twitter's number at all for FUTURE buyers, advertisers, investors, etc. They should show their work and prove their number or investors / advertisers / buyers should lower their valuations to reflect somewhere in the middle. But Twitter hasn't really ever been about the money so the status quo will stay the same. - edit - @blitzar: https://sparktoro.com/blog/sparktoro-followerwonk-joint-twitter-analysis-19-42-of-active-accounts-are-fake-or-spam/ https://sparktoro.com/blog/sparktoro-followerwonk-joint-twit... you can see their methodology here. Despite two different mDAU classifications, both of the datasets from SparkToro and Followerwonk were about 20% mDAU bots.
      - aw1621107 4y ago
        > 3rd party estimations range from 10-20% for mDAU bots Could you provide some sources for these estimates? In particular, I'm curious about the methodology and data sources those third-party sources use.
        hunterb123 4y ago
simiones 4y ago
> If you claim you have such a low number of bots (5%) Twitter has never made any such claim. It's incredible how hard it is for some people to actually read the extremely explicit process they lay out in their SEC filing. Twitter has users. A subset of those users are "monetizable daily active users" - that is, users who log in to Twitter from platforms which are capable of showing them ads, or otherwise making money. A subset of users who would normally be counted as mDAUs may be discovered today to be spam bots, may be suspended for other reasons, or may otherwise turn out not to be monetizable - so they are excluded from future mDAUs. A subet of those users have been counted as mDAUs in the past: this is the subset that is estimated to be 5%. There are plenty of bots on Twitter that are not mDAUs. Some are actually fully legitimate, like the earthquake or weather reporting bots: they are valid Twitter accounts who are bots and who are not counted as mDAUs.
toast0 4y ago
It would be hard for them to be wrong about their numbers, given how subjective they are. Their methodology is pick 100 users from the mDAU every day, then have a human decide botornot. If you wanted to show those numbers were wrong, you might ask to see those specific users and come up with your own botornot score. Or if you thought those weren't inaccurate, but maybe not a representative sample, you might ask for 10,000 users counted as mDAU on a specific day (or each day) and botornot them and see if that matches the 100 sample. If you didn't trust twitter's sampling, you could ask for the full list of users and sample it yourself before botornot. But instead, it seems like they were trying to count botornot for the full population without considering if they were in mDAU or not, which makes it a flawed comparison. The counterclaim says mDAU isn't the appropriate measure for some of the alternate revenue plans Musk was considering, and that's probably true, but it's not relevant to twitter's current business and if their numbers were accurate to the degree they were measured.

pcmoney 4y ago

They literally never claimed this was accurate and even said their methodology could be wrong. Nothing in the contract says anything about bots. They don’t have to prove anything at all.

hunterb123 4y ago
> They literally never claimed this was accurate > and even said their methodology could be wrong Yeah that's my point, their supplied number is unusable for investors, advertisers, etc. Combine that with the bot experiences on Twitter, the 3rd party estimates of 10-20% if looks way off. I don't see why NOW anyone would trust 5%. Diligence has NOW been done and that "methodology could be wrong" looks likely. -- EDIT (post limit) -- > But that's the whole point - Musk waived his right to do due diligence, so he is pointing at this statement being wrong (and not unusable) as his way out. @suresk, that's not a thing, but it is a talking point, but I'm not debating that. I'm saying NOW that the diligence is done and we know it's probably not accurate. See my first post.
- suresk 4y ago
  But that's the whole point - Musk waived his right to do due diligence, so he is pointing at this statement being wrong (and not unusable) as his way out.
- Analemma_ 4y ago
  People keep trying to explain to you why you're wrong and you keep on ignoring them and persisting with your original false statement. The 5% number claimed by Twitter is not the percentage of all accounts that are bots. That has never been the claim. Twitter separates its accounts into "mDAU"s and "other", and they are claiming 5% of the mDAUs are bots. You cannot falsify this claim with "bot experiences on Twitter", because all the bots you're seeing could be correctly classified as "other". Indeed, nobody outside Twitter can falsify this claim because it's a statement about Twitter's internal classification. The only thing that can falsify Twitter's claim is their own internal data, so that's why discovery is being done.
  - hunterb123 4y ago
    I'm not talking about all bots. I'm talking about mDAU bots. You guys are trying to argue against a point I'm not making. 3rd party estimations range from 10-20% for mDAU bots, Twitter's is 5%. Others were defending it by saying "well the number doesn't have to be accurate legally" or "well Musk should have known it's not really accurate". My counter is that weasel legal wording and logic doesn't give much confidence in Twitter's number at all for FUTURE buyers, advertisers, investors, etc. They should show their work and prove their number or investors / advertisers / buyers should lower their valuations to reflect somewhere in the middle. But Twitter hasn't really ever been about the money so the status quo will stay the same. - edit - @blitzar: https://sparktoro.com/blog/sparktoro-followerwonk-joint-twitter-analysis-19-42-of-active-accounts-are-fake-or-spam/ https://sparktoro.com/blog/sparktoro-followerwonk-joint-twit... you can see their methodology here. Despite two different mDAU classifications, both of the datasets from SparkToro and Followerwonk were about 20% mDAU bots.
    - aw1621107 4y ago
      > 3rd party estimations range from 10-20% for mDAU bots Could you provide some sources for these estimates? In particular, I'm curious about the methodology and data sources those third-party sources use.
      - hunterb123 4y ago

simiones 4y ago

> If you claim you have such a low number of bots (5%) Twitter has never made any such claim. It's incredible how hard it is for some people to actually read the extremely explicit process they lay out in their SEC filing. Twitter has users. A subset of those users are "monetizable daily active users" - that is, users who log in to Twitter from platforms which are capable of showing them ads, or otherwise making money. A subset of users who would normally be counted as mDAUs may be discovered today to be spam bots, may be suspended for other reasons, or may otherwise turn out not to be monetizable - so they are excluded from future mDAUs. A subet of those users have been counted as mDAUs in the past: this is the subset that is estimated to be 5%. There are plenty of bots on Twitter that are not mDAUs. Some are actually fully legitimate, like the earthquake or weather reporting bots: they are valid Twitter accounts who are bots and who are not counted as mDAUs.

toast0 4y ago

It would be hard for them to be wrong about their numbers, given how subjective they are. Their methodology is pick 100 users from the mDAU every day, then have a human decide botornot. If you wanted to show those numbers were wrong, you might ask to see those specific users and come up with your own botornot score. Or if you thought those weren't inaccurate, but maybe not a representative sample, you might ask for 10,000 users counted as mDAU on a specific day (or each day) and botornot them and see if that matches the 100 sample. If you didn't trust twitter's sampling, you could ask for the full list of users and sample it yourself before botornot. But instead, it seems like they were trying to count botornot for the full population without considering if they were in mDAU or not, which makes it a flawed comparison. The counterclaim says mDAU isn't the appropriate measure for some of the alternate revenue plans Musk was considering, and that's probably true, but it's not relevant to twitter's current business and if their numbers were accurate to the degree they were measured.