Twitter wants you to build a bot. Me too

For all the controversy about bots above Twitter, most recently due to Elon Musk’s canceled acquisition plan, it seems reasonable to expect the company to close access to the platform and move away from automation. Twitter Inc. is doing the opposite, instead encouraging people to tap into its vast and open trove of data and build software around it.

In fact, it has a manual Online – How to create Twitter robot with v2 of Twitter API – guide you through the process step by step. So that’s what Bloomberg Opinion did. We successfully ran an automated account that retweeted trends, gained a few dozen followers and attracted thousands of visitors. Here’s what we learned about the process and how engagement works on Twitter when an automated system is in charge.

Musk’s assertion that there are far more bots on the system than Twitter has revealed (it claims the number is less than 5%) prompted the billionaire last week to cancel his $44 billion buyout plan. . The company has denied the allegation and said it will fight for the acquisition to be enforced.

Although Twitter was aware of our bot (I told them), the company was not involved in this process nor was it authorized to handle the account. The aim is to mimic how a spammer might set up a bot to spread misinformation or advertise cryptocurrencies, two of the platform’s most common and undesirable uses.

The jargon can be a little intimidating to the uninitiated. Bot is a are from tend to evoke strong emotions. However, it simply means a system that runs with some degree of automation – they are neither good nor bad. API stands for application programming interface and functions as a guide map to access information about users, tweets, images and a lot of statistics connected to them.

A moderately experienced developer should be able to set up a bot in a few hours using any of the major programming languages ​​such as Java, JavaScript, Ruby, R or C. I chose Python and instead of getting started from the ground up using an existing library of free code called Tweepy that makes the process very easy. Some of this code may be old and won’t work with recent updates to the API, but reading the documentation should solve most bug.

The first step in setting up the bot is to open a Twitter account, then sign up for developer access. You can do that with your existing handle if you like. A nefarious actor can do so anonymously by first opening a free show Google account, and then use it to log into Twitter. One phone number is requested only once, for accuracyand there are many online services available for free Internet phone number to help maintain anonymity. However, this is not guaranteed as many internet protocol numbers are blocked from receiving confirmation text messages.

Getting approved as a developer is much easier than expected. There was a series of questions asking what you would use the bot for and whether you would share data with governments. You even have to write down a brief explanation of your project. This process suggests that a human would consider register and reply within hours or days. In fact, it is automatic. Registration and approval takes less than an hour.

Twitter later explained that the quick access was intentional, not a mistake. New developers are included in a low level, called Essentials, which has a limit on the number of times the bot can connect to its system, post tweets and retweets, add users, etc. Its strategy is to engage family Everyone as quickly as possible so they can try out the platform and learn how to use it, while maintaining limits to minimize abuse. If you want deeper access, you. Twitter also requires developers to voluntarily identify an account run by the bot, but it doesn’t enforce it. Since many bots out there don’t seem to have one, neither do I.

Not wanting to trigger Twitter’s algorithms, we chose a generic name and profile that is not associated with any individual accounts or Bloomberg. You can find it @BotofBots2022. After all, crypto bots and spammers don’t use well-known names. We followed @elonmusk, @bitcoin, @CoinMarketCap and @CoinBase to see what other accounts could be attracted we (none of them track back). That association doesn’t seem to do anything to the bot’s follower count.

One what may surprise many Twitter users is the amount of data included in a single tweet. While the general public will see basic content like who posted it, the tweet text, any images or links, and the number of likes and retweets it gets, there’s a lot of information associated with the post. there. This includes how many people follow that user and how many followers. Information about language, location, and time zone, as well as whether tweets are retained in certain countries (for copyright or other reasons), including a web link to sensitive information or if translation is enabled.

This bot is so simple, dare I say boring. On the first test run, it looked for tweets with the hashtag #bot and retweeted. It is set to retweet only a few times per hour for a period of several hours. A handful of accounts started following bots and they appeared to be human (but Who know).

After a few days of playing around, it’s time switch tactic. One of the coolest features on Twitter is the data it aggregates according to trends. The API allows developers to download a list of hottest topics by location. You can set it to worldwide or Home page in languages ​​like USA or Paris. So the Python program was tweaked to take this list, randomly select one of the topics, find a post on that topic, and post back. This gained a few more followers, but not many.

Then we burn after. The bot retweeted 325 times in less than 30 minutes. Sometimes it was stopped with an automatic message from Twitter’s server saying it exceeded the daily limit. So I put it to sleep for a few hours before starting over. It then managed 385 retweets in 26 minutes before being halted again by cap. Overall, the small bot managed 3,200 retweets within 24 hours and over 7,000 within a few days. As the rate of retweets increased, so did the engagement, up to 40 followers. That may not seem like much, but bots just send retweets and yes Nothing clearly interested in its Twitter profile. At the time of writing, it has accumulated over 15,000 profile hits since its inception.

According to Twitter, the two topics that attract the most attention on its platform are cryptocurrencies and erotic, which helps explain why so many bots are built around these two themes. We have made the choice not to add fuel to the flames of misinformation, especially in crypto, and so just stick with existing trends. But even in that more general field, it quickly becomes apparent that a large amount of posts is an easy way to grow followers, which is important for pushing any (false) information. ) you provide.

It would be extremely easy to configure this bot to tweet (instead of retweet) crypto-injection posts, scam ads, or link distribution. And with the generous limits on which an Essentials developer account can be opened within an hour, there’s plenty of room for the bad guys to jump in and use the platform. Spam before being started. To protect itself, Twitter has both humans and automated systems in place to prevent such malicious activities. It’s likely that our bot was doing too well to trigger any alerts. Executives don’t like to share exactly how the company spots the bad guys. It’s a non-stop game by Whac-A-Mole, and Twitter wants to avoid the system’s game-learning enemies.

The biggest takeaway from building a Twitter bot is understanding how easy it is and how powerful a tool it can be. Not all automated processes are bad, nor are all human accounts good. But the automation of the platform makes it incredibly easy to ramp up use – tweets on steroids. There are tons of great bots out there – post pictures of cats, collate content, provide earthquake information, or share an insider joke. And there are many more maligns (we won’t name them).

Over time, and under new management if Musk ends up owning the company, the world and Twitter itself will likely rethink its approach to automation. That would be a pity. There is no doubt that Twitter bots are being used for malicious purposes, yet there are many instances where open access to Twitter data serves as a public good – including tracking the flow of misinformation by human-run accounts. Setting curbs for bots can be guaranteed, but removing them entirely would be overkill.

Tim Culpan is a technology columnist for Bloomberg Opinion in Asia. Previously, he was a technology reporter for Bloomberg News.