Twitter wants to banish the bots in its network and is taking action. It seems that their actions are working, it ensures that less than 5% of the active accounts are false or spam.

We are accustomed to seeing news about the discovery of Twitter accounts full of bots. For example, Newsweek reported that nearly half of the followers of Donald Trump were false. And a study published last year by the universities of Indiana and Southern California he noted that more than 15% of the accounts on Twitter -about 48 million — were in fact bots, and as you gathered CNBC.

however, the data collected by Twitter are very different. “According to our internal review of a sample of accounts, we estimate that less than 5% of our monthly active users (MAU) are fake accounts or spam”, the company said. “Currently, we detected approximately 523.000 suspicious logins each day, that we believe that are produced through automation. In December of 2017, our systems identified and challenged by an average 6.4 million of accounts suspected to the week,” he adds.

So, it seems that Twitter has been serious in the fight against the armies of bots. “The content that violates our rules concerning accounts, automatic spam can have a highly negative user experience. And for a long time we have taken substantive measures to stop the flow”, he asserts.

Actions to stop the bots

“To be clear: Twitter prohibits any attempt to use the automation with the goal of publishing or disseminating spam,, and such behavior may result in the application of coercive measures,” says Yoel Roth, of the division of policy API and confidence of the product of the company, in a post in which he explains the measures that are being put in place in order to catch the bots on your network. “In January, we announced that, as part of our efforts of quality of the information, we would do changes in TweetDeck and on the Twitter API to limit the ability of users to perform coordinated actions on multiple accounts”, specified by Roth.

microblogging network believes that “the most effective way to fight against bots suspects is to stop them before they start. To do this, we have created ssystems to identify attempts by suspects to log on to Twitter, which includes evidence that a login may be automated or programmed.” It points out that much of this advocacy work is done through machine learning and automated processes in their back-end. And claims that it has significantly improved their automatic detection of spam and bots. “We are investing in systems to stop inappropriate content in its origin, if the point of origin corresponds with an actor’s fraudulent known. However, the use of proxy servers, VPNS, and other forms of authentication, especially outside the united States, can obscure the true origin of the traffic” points.

The company also emphasizes that is improving the way to detect and group accounts created by a single entity or a single suspicious source. “we Use these techniques to stop more than 5.7 million of spam from a source in only a week,” he said. Also, is moving forward in the discovery of patterns of activity is not human based on signals such as the frequency and the time of the tweets, retweets, likes, and follow-up of accounts. It has also created models to detect if an activity is automated.

on the other hand, is working to identify accounts whose security may have been compromised, in order to prevent them from being exploited by actors that are fraudulent. “We are creating systems to detect when the login activity is not consistent with the typical behavior of a user, for help that the compromised accounts will be under the control of their owners”, review.

in Addition, Twitter is verifying the contents of a suspect. So, the accounts and the content detected by their systems are subject to a series of actions and limitations, as being in mode ‘read only’ pending authentication, to limit the scope of tweets on the basis of their origin suspect or of poor quality, the removal of the associated content or even the suspension of the account.

Also, is working on the identification and adoption of measures against the third-party applications that violate its policies of development, including bots and automatic applications. “Since June of 2017, we have suspended more than 117.000 malicious applications for abusing our API, collectively responsible for more than 1,500 millions of tweets that are of low quality,” he said.

Twitter acknowledge that any automated system to detect spam or bots, can give rise to ‘false positives’, by ensuring that often give users the opportunity to verify that they are legitimate before you suspend your account. “During this window, the accounts may still appear on Twitter and through our public API, despite not being able to create new tweets. We also limit the visibility of these accounts and their contents both in searches and in trends”, he says.

finally, has improved the verification via mobile phone. “When we detect suspicious activity from an account, we may require that user to verify their phone number to regain access. As spammers have adapted their techniques, we discovered that not all phone numbers are equally reliable.

We have improved our reputation system phones to identify operators and suspicious numbers and prevent its repeated use to overcome the challenges of verification,” notes the company.

How to avoid cancellations?

Twitter made a series of recommendations to prevent their systems from getting confused our accounts with bots, especially when using applications that allow you to manage multiple accounts. In the first place, advised that you do not publish simultaneously contained identical or substantially similar across multiple accounts. “As an alternative to publish identical content, you can retweet content of an account from the other accounts that you want to share that publication”, says the company. He warns, however, that can only be done from a small number of accounts directly controlled by the user, because the retweeting high-volume, automated is not permitted by the rules of automation of the platform and runs the risk of being subject to coercive measures. It also recommends that you do not make the simultaneous actions -likes, retweets or follow a specific account — via several accounts.

in Addition, he notes that you are not allowed to use any form of automation or programming to publish content identical or substantially similar to, as well as to perform actions on many accounts, regardless of whether the user has created or directly control those accounts. For example, are banned applications that coordinate the activity in multiple accounts to post tweets at the same time with a specific hashtag in order to make the item display as a trend.

Exceptions of public service

Twitter knows that it has become a powerful tool of communication in certain emergency situations, so in some cases, that allows the automation of tweets.

“while we continue to allow the posting of external information -such as weather alerts or RSS feeds — through automation, you should only publish this content in an account controlled by the user. As the sole exception to this rule, the applications that transmit or share weather, emergency or other public service announcements of broad community interest, as warnings of an earthquake or tsunami, can publish this content on multiple accounts that have authorized a application”, indicates the company.

Is it a good idea to buy followers?

Twitter prohibits the purchase of followers. Even warns of possible risks of the apps that offer followers for free. But it is well known that there are companies that offer buy followers.

These companies argue that users that offer are real, not bots. In fact, recognize that it is very possible that Twitter take action against an account full of bots. To be real users, it is more difficult for the network the automatically identified as fraudulent accounts. However, the interaction of these users also don’t follow patterns ‘real’, so that these maneuvers can also be detected.

on the other hand, almost all of us have received a tweet from ‘follow me and I will follow you’. offers the follow-up to mutual with the aim of artificially inflating the number of followers. These accounts are gathered thousands of followers ‘actual’ and after they are sold. In case of being detected by Twitter, it is very likely that you will end up being cancelled.

The companies that sell followers to ensure that they go to them politicians, athletes, artists, journalists, means of communication, etc Also come to this type of companies those users whose accounts have been cancelled, with the purpose of recovering immediately the number of followers had. However, it is a practice that is risky, because you run the risk of the platform to detect the fraud and try to cancel the new account.