Elon Musk knew it since the very beginning, and to be honest you knew it too. The bots and spam accounts on Twitter make up more than 5%. They are in greater numbers on Twitter than they are on any other social media platform. This is finally backed up with facts – the company that Elon Musk paid to do the analysis confirmed it.
The company Cyabra held its first official interview a couple of days ago and it revealed its findings. According to their report, Twitter has a hot spam accounts pool of about 11%. This hints that Elon Musk will have a lot of work to do if he wants to get rid of them completely. Now that he agreed to proceed with the acquisition of Twitter for the agreed $44 billion, it will be inevitable to face the bot battle.
What does Elon Musk’s Bot Analysis Show? – Will He Be Able To Eliminate Them When He Takes Over?
Dan Brahmy is the CEO of the analysis company Cyabra. Recently, in an interview with CNN, he revealed that he performed this type of analysis on other social media as well. He assessed the company’s rival social media platforms and found out that Twitter is in a far worse situation currently. Dan Brahmy refused to name any of the other companies he assessed and to reveal the exact numbers he got, but he confirmed he can back up the claim with numbers.
“We have a number for all the social media platforms because that’s our job.” – Brahmy told CNN. He even claimed that his company had some data on Twitter even before Elon Musk hired them to do the analysis.
The bots and spam accounts on Twitter were the main reason why the acquisition of the company took many months. Elon Musk even tried to get out of the deal on that fact. Three months into negotiations on the acquisition process, Musk decided to terminate the deal. Even though one of his main reasons for buying the company was to address the bots on the platform, all of a sudden he gave up before even starting and decided to back out.
Twitter firmly stands behind its claim that the bots and the spam accounts on the platform don’t make more than 5%.
Parag Agrawal, Twitter’s CEO reassured his Twitter followers of this fact over and over in the past years. Most recently in May, he said that “spam harms the experience for real people on Twitter.” Adding that “as such, we are strongly incentivized to detect and remove as much spam as we possibly can, every single day.”
Twitter filed a lawsuit against Elon Musk after he revealed that he will terminate the deal for acquisition. They accused the billionaire of only using the bots and spam accounts as an excuse to get out of the deal. According to them, Elon Musk has buyer’s remorse. As a result, he tried to terminate the deal.
However, after a long doubting, questioning, and thinking process, Elon Musk decided to go on with the deal. Last week, he contacted Twitter and publicly accepted the terms of the deal, and decided to follow through. The planned trial is on hold as the judge gave both sides until October 28 to close the deal.
Since April, when Elon Musk informed the public of his intentions for Twitter, there has been an unstoppable supply of information on the internet, from unofficial, and official experts. Data scientists, artificial intelligence experts, and misinformation researchers took sides and expressed their beliefs, and showed their results for Twitter.
Cyabra is a small research and analysis company in Israel that by some chance is the company of choice for Musk. They have about 40 employees and numerous clients from all around the world, including the US. Even though Cyabra gave a hint on the real number of bots on the platform, now when Musk is about to close the deal – we might never find out the real numbers. Most probably this information will remain closed in Elon Musk’s drawers until solved. And that is questionable as well. Will he keep up to his word and try to solve the bot issue?
Additional to all this, people who work in the data analysis industry warn that the estimation of spam or fake accounts can be very tricky. On top of that, they warn that this type of work could be a very subjective task. While companies and people are coming out with exact numbers about spam accounts, they underestimate the complexity of the issue.
Botometer is another service that has an association with Elon Musk when it comes to bot analysis. Indiana University Network Science Institute and the Center for Complex Networks and Systems Research are the co-founders of the tool. They released a statement that when working on Twitter’s bot issue, they faced a massive challenge. They struggled to define the word ‘bot’. What do we consider a spam account? How can we tell with certainty that some account is a ‘bot’? These were questions that complicate the matter to an extent that it becomes philosophical.
Back to Brahmy’s claims and his company Cyabra’s analysis. Brahmy states that Cyabra is mainly using a machine learning algorithm to monitor the accounts’ activity and with that determine whether behind the ‘wheel’ of the account we may expect to find a human being or a computer. According to him, the estimation of the spam accounts has an accuracy of 80%.
The reason behind this is that there are so many users on Twitter that are not as active as daily users, but with their presence on the platform, they contribute to the analysis. He took the example of some older lady that has an account on Twitter. She would usually not post much, or not write lengthy tweets, many times her tweets and re-tweets will be outdated. In addition, she would make some grammar or spelling mistakes… A typical bot activity. However, that doesn’t put her in the bot ‘basket’.
Twitter’s lawyers state that none of the analyses the billionaire made, or paid to be made, can be taken into consideration when speaking about bots on the platform. Despite the fact that one of the companies that Elon Musk hired claimed that their results are with 90% accuracy. Twitter scraps Elon Musk’s claims that the amount of spam bots is “wildly higher than the Twitter estimate”.
In May, Cyabra made ‘free’ analysis and published them on their website, estimating that Twitter has about 13.7% spam bots. Seeing these results, Musk requested new data analysis from the company, that will be more accurate and reliable. To achieve this, Elon provided the data he obtained from Twitter directly.
Musk used the ‘paid’ analysis in the court filings, stating that the accuracy of the numbers is about 80%. According to that data, the spam bots on Twitter make up about 11% of the total user base.
“The request was precise: Spam and bot accounts. Tell us the number, tell us the methodologies and tell us the confidence level.” Cyabra worked on their own, not seeking help or using data from any other company, and did the calculation as precisely as the conditions allowed them. They did not specify the amount that they charged Elon Musk for the analysis. However, the fact that they took weeks of data gathering and calculations to do it – suggests that it is a ‘salty’ one.
Another important detail to mention when speaking about this issue is the mDAU and total Twitter users. The difference between the two is massive. For example, mDAU, or monetizable daily active users are far less than the total number of accounts on Twitter. The thing is that Cyabra, cannot state with certainty that they monitored only mDAU accounts. When asked about this, Brahmy declined to elaborate. We don’t know what may be the reason that they are not able to distinguish between the two.
Cyabra’s algorithm takes into consideration numerous things when determining if an account is a bot, or not. The accounts the user connects with, account’s location, what are their tweets about, are just some of the guidelines. In addition to this, the analysis includes the frequency of their activity and the hours when the accounts are active.
After a long process, Elon Musk finally agreed to take over the company. Will he be able to get rid of the spam accounts for good? Will he hold up to his word and make it a censorship-free platform? Well, we have the following few months to find out.