==========================
== Neural Market Trends ==
==========================

Why (most) Twitter Bots Suck

AI Machine Learning Python

I’m going to be the first to admit that I use Python to send out Tweets to my followers. I have a few scripts that parse RSS feeds and do retweets on an hourly basis. They work fine but they do get ‘gamed’ occasionally. That’s the problem with automation, isn’t it? Getting gamed can cause all kinds of havoc for your brand and reputation, so you have to be careful. Has this happened to me? Not really, but there has been a few embarrassing retweets and silly parsed advertisements in lieu of good articles.

Why bother with Twitter automation in the first place? Simple, everyone wants to be an ‘influencer’, myself included. Yet using automated methods to gain ’eyeballs’ comes with a price. You end up sacrificing quality for quantity. You end up diluting your brand and losing the signal. In the end you get nothing but noise!

Signal vs Noise

At one time I tested/used @randal_olson’s TwitterFollowBot to increase my follower count. It worked well and I started growing my followers in large clips. The script is pretty simple in logic, it follows people based on a certain hashtag (or followers of a Twitter handle) that you supply and does in about 100 people per run.

The goal here is to get a ‘follow back’ from the people you just followed, then auto mute them. If, after a week or so, they don’t follow you back you run another routine that ‘unfollows’ them and puts them on a black list not to ‘autofollow’ them again.

You run this script every few hours for a week and MY GAWD, does your following list explode! The noise becomes unbearable, even after muting them. You end up with cranks, conspiracy theorists, crypto-currency shills, and bots (most liked Russian bots). Yes, you do get a lot of follow backs but the quality signal of people you should really follow and interact with gets completely lost!

I stopped that experiment a few months ago and started unfollowing the noise. My following count is now below 1,000 but I feel that’s too much. I want to get that number to about 500. Of course, this resulted in my follower count dropping too. There’s a lot of Twitter users that also run ‘you unfollow me so I unfollow you’ scripts too. LOL.

Possible solutions

Just stop it. Stop all the Retweeting, TwitterBot following, and parsing. Instead do one or more of the following:

    - Create a curated list of great links that you filter through. I know that @maoxian has done this over the years and it's invaluable because he puts the time and effort in to filtering out the noise.
    • Write a Python script to parse RSS feeds but write the links to a file so you can review later and tweet accordingly (more signal, less noise)

    • Write a Python script to find ’true’ influencers on Twitter and interact with them personally. Perhaps create a ranking system

    • Something else that I’ll remember after I post this article

I guess lesson here is that we can’t automate the human touch. You can do a lot of the heavy lifting but in the end, it’s us that bring meaning and value to everything we do.

Update

The only Twitter Bots that Don’t Suck are the ones that have a curated reading list. Those require fine control on the part of the end user to come up with a good list that’s constantly refreshed. A simple python script that reads an entire list of articles and tweets them out one by one is probably the right thing to do. Run it like cron job but replace the list every week.

comments powered by Disqus