ICYMI (ADN) bot analysis - May 2023

Richard Martin-Nielsen · June 23, 2023

Introduction

This is a rough post-facto analysis of the behaviour of the ICYMI (ADN) bot. The bot spits out some rudimentary logs as it works (in fact as part of how it processes creating a list of toots to boost, and tracking which of those it has boosted and which it still has to boost) and this stores high-level data about the promoted toots (but not the discarded ones). I use the logs along with data drawn from the server using the rtoot library to slice and dice the data and try to present the data graphically. This is still largely an exploratory rather than explanatory exercise.

Time delay

New analysis this month lets me compare the delay between when a toot is created and when the bot boosts it. The quick conclusion would be that higher scoring toots may be boosted sooner after their initial creation - which makes sense considering how the bot scores toots and considers which to boost.

Scores and time

The bot assigns a score to each toot it sees, then boosts the toots which get the highest score. This can mean that when there aren’t many toots, toots with lower scores might get boosted which otherwise wouldn’t be amplified. (I think this happens on weekends.)

First, a quick look at how the scores range over the month. Not much. There are a few outliers.

But even when I crop out the outliers, there’s not much visible pattern to see, except fewer toots boosted on weekends.

Accounts boosted

Anyone who follows the bot and will see that it tends to boost some accounts a lot. This isn’t very surprising since (a) it doesn’t follow many accounts and (b) even those accounts who do post about ADN topics don’t all consistently use the hashtags which the bot listens for.

Still, a histogram shows that nknews, nukes_of_hazard and Livableworld are most likely to score well and be boosted.

Scores of toots from regularly boosted accounts

Looking at the scores of toots from the most frequently boosted accounts, there is some variation.

Day of week variation

I’ve wondered if there were any weekly cycle in the scores of the toots boosted - with a slight thinking that on the weekends when there is less “official” or “news” traffic, the threshold required to be boosted might drop. There might be slight dip on Saturday and Sunday, but I suspect that my scoring is blind to whether items are popular and interesting due to being topical or not.

On the whole, there are fewer toots boosted on weekends.

Sites referred to

Looking at which sites the toots link to there is greater variation, though nknews.org, linked to the @nknews account, gets almost half the links.

Beyond that, there is a mix of press, NGOs, and other specialised media outlets.

Keyword frequency and topic analysis

A very very basic keyword search was used to mark all boosted toots based on the content of the toot (not any linked site). This was then used to look at which topics are most frequently referred to, and where they may overlap.

More than half all the toots referred to nuclear, followed by Korea, “missile” and other uncategorised toots. This month, “Korea” and “nuclear” together came more frequently than “Korea” and “missile”.

30 of the boosted toots (out of 265 total) didn’t fall into any category.

“nuclear” and “power” appear together in 11 toots and it seems the bot is amplifying discussion related to nuclear power plants.

Comments

This analysis is a very light reworking of the code done for the analysis of data in April 2023. The code has been adjusted slightly and should now be easier to run on an entire month’s worth of data - and I hope to run it again soon with data for June.

In that analysis I had a list of next steps. I’ve really not done much to implement any of those.

One thing I’ve found in my looking at the bot and how it manifests in my timeline is that it is relatively good at picking among sources of toots, but doesn’t try to pick among sources of boosts. So if an account which it follows which it otherwise scores highly boosts several unrelated toots in quick succession, these toots may win the scoring for a while and the bot becomes an echo of the boosting activity of that account.

I have some ideas how to damp down this enthusiastic call and response from the bot but I’ve not started trying to code them.

The other larger but less “code-y” project for the next couple of weeks may be to clean the code and make it a more obviously independent fork of Matt Hodges’ original mastodon digest project.

Twitter, Facebook