Notes on the Banning mechanism

Gresham’s law, is an observation in economics that “bad money drives out good.”

The wider interpretation of that principle helps us understand how tolerating bad actors in an ecosystem will drive out the good actors and ultimately harm everybody. Every dollar given to a cheater is essentially a dollar taken away from a legitimate participant. Plus, one cheater can create hundreds of accounts and thus drive away all the real users. Thus, banning becomes essential to drive away the cheaters.

In Hashtagger’s ecosystem, we want to have micro-influencers (normal users who have 30–30,000 followers ) who actually pay attention to the project they are tweeting about and by doing so share it with their friends and followers and get paid by the engagement they earn legitimately due to their work in creating the tweet and growing their network.

However, some people see that we’re paying as per the engagement and use that information to fake engagement by various dubious practices. This is the reason why we have implemented strict algorithmic banning mechanisms for those who’re engaging in such fake engagement activities.

Some such dubious practices are:
1. Using bots etc. to drive up their tweet’s engagement

2. Paying people to boost up their engagement.

3. Making new alternate/fake accounts to boost up their own engagement.

4. Forming groups where people just exchange likes for each others tweets often with their new alternate/fake accounts.
etc.

While these tactics may drive up likes, comments and retweets, they also leave behind a clear trace of being unnatural.

Since cheaters only focuses on the engagements that gives them the payment, they don’t pay attention to other parameters such as their followers, post’s views, posts’s clicks, likes and other engagements on comments, their average engagement etc.

So, a normal person’s post will have an engagement similar to or 2–5 times higher than their average engagement, but a cheater will have 10–50 times higher engagement than the average for his other posts.

Normal posts may have something 1:10 ratio for engagement and views, but a cheater will have a much higher ratio for their tweets.

Some examples are shown below:

1. Fake accounts:

Above, we can see that a person has just made multiple accounts by changing the numbers at the end and has used it to boost their own engagement. The algorithms will catch this and even much more sophisticated fake engagements.

2. Fake engagement

Some people, like the one in the above tweet, will have a completely disproportionate number of followers and engagements, either through bots or spending money on this, or by using fake accounts.

3. bot engagement (fake)

Some people used pretty visible bots such as coinkit for buying engagement. Some used other less visible bots or telegram groups where members created multiple accounts and liked each other’s tweets. We will detect the signature on such tweets to be unusual and will ban them.

Do note that if your tweet goes viral for real and your engagement surpasses your follower count by quite a bit, you won’t need to worry about it because your engagement:view ratio will still look good. Only those who’re using shady practices will be caught by the algorithm. The offending tweet shown above is definitely not a ‘viral’ tweet.

We have started to implement the ban from our second campaign onwards and here are some of the metrics we considered at that time through our algorithm (the list of metrics we consider for this is likely to evolve in the future as the cheaters get more sophisticated):

For those who are statistics oriented, here is a detailed list of all the tweets which were banned during the second campaign. As we move forward our criteria will further evolve.
https://docs.google.com/spreadsheets/d/1Vj3hxgqMcqCoGQ8cVNIeF_gLNDXjXf1DdkjYR9HaEeU/edit?usp=sharing

In addition one can also be banned for the following:
Posting any explicit or nsfw content in the pool.

• Posting religious content in the pool.
• Posting political content in the pool.
• Posting any hate speech in the pool.
• Spreading misinformation.
• Using bots and/or hiring people to artificially inflate one’s tweet engagement.

To sum it up, all accounts detected to indulge in the shady practices will be banned. For our first round of bans, we’ve published the data in a spreadsheet above to show the details to everyone and make the ban last only for one campaign so that people can correct their behaviour.

If you got a temporary ban this time, spend time in properly learning about the crypto ecosystem and sharing your knowledge with people and organically growing your Twitter account. This will be much more beneficial for you in the long run and we’ll be happy to help you with that journey and reward you with our various campaigns.

Those who are intent on cheating the system will continue to evolve though. The following was a rather interesting cheating confession message we received in our telegram:

Our algorithm will continue to evolve too and cheating will become tougher and tougher on the platform. So we do hope people spend their time trying to genuinely grow their reach and influence on social media and we’ll be most happy to help them do that.