Hey guys, it's sasha, the battle between elon musk and twitter just got a lot hotter and i am beginning to run out of popcorn supplies, because in the last few hours twitter, ceo parag agrawal has tried to defend the company from accusations of faking, their user numbers. Elon musk went to town and absolutely destroyed twitter ceo on twitter and the stock market is now thinking. The deal is more than 50 percent likely to not happen than it is to happen. So what the heck just happened, and where is all this going? Well, i'm gon na go and break it all down for you.

The story starts yesterday at half past nine. In the morning san francisco time, when parag agrawal decided to go and defend twitter's sec filings that have claimed that the platform has less than five percent bots. On it, because twitter has been making the exact same claim in every quarterly report, since q2 2014 here is the q2 2014 earnings presentation, and the wording here is pretty much identical to the wording used today saying that false and spam accounts make up less than five Percent of the active users back then it was five percent of the monthly active users. Today is five percent of the monetized daily active users, but i guess it's all the same thing, because the point is, it is always less than five percent.

I am guessing twitter's. Lawyers told the twitter board that five percent is this magic number where, if your spam and bot accounts are less than that, then advertisers won't really care and the sec will probably be fine with the metrics that you report to them, because it is somewhat strange that, If you go and do this analysis every quarter for eight years in a row that you always get exactly the same result of it just being less than five percent instead of maybe i don't know three percent or one percent or six percent or some other number. So, a few days ago, elon musk put the twitter acquisition on hold saying that five percent number seems a little bit low, because anyone who uses twitter can see the huge volume of bots on the platform and parag put together a carefully worded thread to debunk elon Musk now in this thread he said: let's talk about spam and let's do it with the benefit of data fact and context. So, to start with, he took a digger deal on insinuating that elon's point of view and some other people that maybe agree with him is bereft of data facts and context.

You know this one is going to be good when punches are thrown from the get-go. Now i'm not gon na go and read out the entire thread, because it is long most of it is boring as hell, and it is full of garbage. Parak spends a few tweets saying that spam is bad. Blah blah blah feel free to go and read the whole thing yourself.

If you want the full picture, but then he gets to the numbers, he says we suspend over half a million spam accounts every day, usually before any of you even see them on twitter. We also lock millions of accounts each week that we suspect may be spam if they can't pass human verification challenges, captures phone verification, etc, and here parag shoots himself in the foot without realizing it, by trying to showcase how great twitter's bot detection system is, because if You look at twitter's q1 report. It says that twitter had 229 million monetizable daily active users, and that is up 15.9 percent since last year and just so we're accurate here's. The q1 report from last year, which says that it had 199 million monetizable daily active users except wait.

This number is actually wrong because remember, twitter admitted that they've been double counting their accounts for the last three years, so they can't even add up the accounts, let alone identify which out of them might be bots. So the actual number according to twitter in q1, was 197.6 million, so from q1 2021 to q1, 2022 twitter added 31.4 million monetizable daily active users. And if we're going divide that by 365 days, that's about 86 000 new users every day, but parag just said that they delete over 500 000 spam accounts every day and on top of that, a bunch of accounts, no number provided are also blocked. So parag was trying to showcase how epic twitter's bot detection system is, how many accounts it identifies, except.

He just admitted that 85 of all new accounts joining twitter at the moment are bots. Five out of six new accounts are identified as spam and if they don't identify some of the others, that ratio is even worse and so, unfortunately, instead of showing twitter's technological prowess, the ceo of twitter just admitted that the company absolutely sucks balls at stopping those bots From signing up on the platform in the first place, there are plenty of examples of other large tech companies, including social media, that seem to be far better able at doing this. But you know this involves actually trying to build solutions, and it is far easier. Instead of doing work and finding a way to go and say that finding bots is really hard, and you know we did our best - maybe a bit of investment in the signing up process so that you don't have to then identify and delete 500 000 bots.

Every day could help just a tip there, and here is where this screw-up gets even worse, because i am guessing even parag would admit that twitter's bot, detection, algorithms are not perfect. Actually, i am not guessing at all. Parag literally says that in one of the tweets in this very thread, but let's say that twitter is really really good at identifying bots. Let's give them the benefit of the doubt.

Let's say that they catch 90 percent of all the bots out there, because they don't bother trying to stop them from registering in the first place. If they are that good at catching them and by the way, i don't think for a second that they are because that is a ridiculously high bar. But let's say that means that there's only 10 of the bots that they don't catch, and that means that if they catch 500 000 every day they miss another. Roughly 50 000, which is more than half of the 86 000 new accounts per day.

The twitter say that they got over the last year and if every day you're adding more spam accounts to the platform than real ones, then how does the total of the spam accounts after this has been going on for about eight years? Every single quarter stay under five percent. A really interesting quandary parag is very careful with his words. He says we have estimated and the word estimated is crucial because it is roughly equivalent to saying we guessed. You can estimate something without basing it on hard specific analysis and data by inference or having some very loose, not very closely related analysis.

Elon musk said that twitter only runs a tiny sample of 100 accounts a month to do these checks and for a multi-billion dollar company. That seems absurd and ridiculous, because this is in no way statistically significant or representative and is completely dismal as a way of doing any meaningful checks. Apparently, twitter's lawyers then called elon musk and were angry that he had the audacity to disclose the fact that they only check a hundred accounts a month, because you know that's sensitive information. I've got a tip for twitter's lawyers here, a free one, maybe try a little harder to make twitter actually do proper statistics that will be actually meaningful recognizable before you know, publishing that data to the sec instead of being angry, when someone points out that your data And you're an analyst completely suck balls, so parrok said our estimate is based on multiple human reviews in replicate of thousands of accounts that are sampled at random consistently over time from accounts.

We count as monthly daily active users. We do this every quarter and we have been doing this for many years - very careful choice of words there. It references thousands of accounts and multiple human reviews, but it does not say if this is maybe per month per quarter per whatever. It doesn't give any sort of time based reference point and because it does not, it really does make sense.

What elon said? Because if you add up loads of quarters of 100 accounts each then yeah you do get thousands? The issue is, if you have 229 million accounts on the platform, then a random sample of a hundred or even thousands, is incredibly low and will give you completely meaningless results. Let's say that thousands means a few thousand. Let's give twitter the benefit of the doubt again and let's call it five thousand paragraph said our actual internal estimates for the last four quarters were all well under five percent based on the methodology outlined above. He also says the error margins in our estimates, give us confidence in our public statements each quarter, oh confidence.

Well, let's look at it in a bit more detail, so you want to say that we think the number is five percent and we want a 95 percent confidence interval around it saying that is between four and six percent. You know to get it roughly in the right ballpark. If you want to do that, then you will need to sample 192 000 out of the 229 million accounts, not just five. If the proportion of bots is much lower than five percent, then that sample size would have to increase considerably to get the same level of confidence so yeah.

Unfortunately, the ceo of twitter and whoever the waste of space data analysts work. There clearly do not understand even the most basic statistics that get taught in school. Then parag decides to do a bit more defending. He says.

Unfortunately, we don't believe that this specific estimation can be performed externally, given the critical need to use both public and private information, which we can't share externally, it's not even possible to know which accounts are counted as monetizable daily active users on any given day so much Wrong with that tweet, first up sure twitter has a lot more data about the users than what i can see. When i look at an account. Twitter knows everything about the behavior. What people look at what they click on, how long they spend on each tweet time stamps ip addresses blah blah blah blah blah.

I get it so in theory, that should mean that they are way better than me at identifying bots right. Well, they have way more information than a bunch of massively overpaid analysts, except i went and did a little experiment. I just posted this on twitter a few minutes ago. Let's see how long it takes to get a bot comment, if i mention trustwallet, bitcoin, luna save moon, dosh coin or flocky inu, especially if i say that i lost my coins and i need help.

I seeded some bait with all the popular crypto, buzzwords and sure enough within about five seconds. I got a reply from trust, crypto wallet, username, joya loads of numbers afterwards, and it says we are sincerely for experiencing this amazing grammar. There kindly get in touch with us via dm by clicking on the below button, and we will be able to help. Please disregard any other messages.

There are scam. This is so epic because i don't have an army of analysts or supercomputers, and i can identify robots by them replying with this sort of message within seconds trying to scam people out of money and that bot literally posts. The exact same thing over and over and over to many people all over twitter every few seconds you know that would be probably extremely difficult to go and identify the truth is twitter, just don't care identifying. This sort of behavior would be ridiculously easy, but the problem is that deleting bots is bad for business, because advertisers pay for brand marketing based on impressions and the more monetizable daily active users.

You have the more money you make from those ads and a couple of minutes later i got another one. This one is terra usd powered by luna. They can also help me if i go and send them a private message. How incredibly useful so perrick is saying that it is not possible to know which accounts are counted as monthly daily active users on any given day, and that means that perhaps they do identify some of these spam robots.

But they don't count them towards a metric, which is a really weird thing to infer, because it kind of means that perhaps they know that they have these robots. But you know as long as they don't count them in that special figure that they report it's. Okay. It's okay, that they scam people and steal people's money.

That sounds like a somewhat stupid idea, and here is an amazing thing. I do not have access to twitter's amazing data. I just went and found two spam bots in about five seconds worth of work, and it seems that elon musk agrees with me on the fact that there is a few too many of them because he went and posted a turd underneath the whole parags thread. A very succinct summary of all the information inside it that i completely agree with.

There are many studies out there that place a spam volume on twitter at more like 20 and elon. Musk just spoke on the all-in podcast a few hours ago and he said that the number is incredibly important. You say the house has less than five percent termites. That's that's an acceptable number, but if it turns out it is 90 termites.

That's uh, not okay! You know it's not the same house if twitter is claiming that they have less than five percent of bots, but in reality it is substantially higher. This is a material difference to the value of twitter as a company, because it implies that advertisers are currently wasting a whole lot of money on advertising to bots and paying money for it, and the real user numbers on twitter are considerably lower. Once again, twitter is the most actively traded stock in pre-market this morning, and the share price is now at around 36, as i'm recording this video given elon's takeover bid is at 54.20. That is a 33 discount off the price, which implies that most analysts now don't think the deal will even go through at least not at that price, because if elon wants to pull out of the deal - and he has enough smart lawyers to figure out how exactly To do it, maybe he'll have to forfeit the one billion dollar fee.

Maybe he won't if he uses the twitter numbers as the reason, because if he did, twitter would have to go and sue him to go and force him to buy, and presumably they might have to then disclose their exact methodology for calculating that number in court. During discovery, i have a funny feeling that they really don't want to go there, but if elon walked away, twitter share price would most likely dump all the way to at least 20 to 25, because every other tech stock has had a massive sell-off during this period. While the twitter saga has been unfolding and 36 dollars is much closer to 25 than it is to 54., i'm going to go and load up on some more popcorn to see what happens, and if you like this video, please don't forget to smash the like button. For the youtube algorithm, thank you so much for watching and i'll see you guys later.

.