< back

Creating Topic Newsletters From Mastodon

May 12, 2023
Toot LinkedIn

So far, fedistats has been a journey for me. The original idea was to compute basic statistics and see what interesting things might pop up. But I also wanted to see whether this could be turned into something that is actually useful for people. So I kept using it, coming back to it day by day to see what new trends might be popping up.

And then artificial intelligence really blew up in March. Looking back I cannot even remember how it all started. Something about Meta releasing LLaMa for research purposes and then the inevitable happened, the weights got leaked and people started to get their hands on it, and someone wrote a piece of highly optimized code that meant that you could run the model on your own computer. Suddenly it felt like AI was here. And then OpenAI released GPT-3.5, the biggest language model yet and things just kept accelerating from there.

I had seen early on that the #ai hashtag surfaced a lot of the news, often within hours. Every time I checked back I was seeing something new, often before I would learn about it from other sources.

An Automated News Feed For AI

So I asked myself, wouldn't it be cool if I could track all the new articles that had enough mentions to make it into the top 10? I built a first prototype and it has been running since mid of April collecting already more than 300 articles, on average 11 per day.

The three latest headlines from the article feed as of May 12, 2023

After a few days I thought what if I sent myself all new articles? Would I quickly tire of these or is the content actually interesting enough. So I added another bit to fedistats's ever growing list of jobs to send me a DM on mastodon every time it found a new article that had enough mentions to make it into the top 10.

I found that it really helped with my "AI FOMO." So I decided to open it up a bit more and now these aren't DMs but private messages all followers of fedistats account can see. If you're already on mastodon, follow fedistats and you can see for yourself.

But Who Would Use This?

At this point it dawned on me that the people who would find this useful are not the people who are already on Mastodon. If you are, you're probably enjoying the whole social media experience, and you'd have your ways of finding the content you find interesting.

End of April I was preparing a presentation for a client on the latest developments in AI, I realized that so much of the content that was in my head was actually something I found through fedistats. So something of value was there, but how could I best deliver it?

To make it more accessible, I created an RSS feed. Given everything that I had, this was very easy to do, but it seems nobody subscribed to it. I guess RSS is dead, after all... You can still subscribe to it here.

But everyone has an email, and companies like substack show that engaging with content through email is still something that works for many people.

So last week (wow, has it only been a week), I added a feature where you can subscribe to fedistats AI topic and get a daily newsletter with all the latest articles found in the last 24 hours.

This is how it looks like and even though I'm already consuming the content via mastodon notification, I still find myself going through the list and finding interesting stuff I had missed.

So this is where you come in! Are you interested in AI but cannot keep up with everything that is happening? Want to figure out what I'm working on? If you want, give it a try and subscribe. The feed follows a couple of AI related hashtags and looks at content from a number of reputable tech news sites. You can also configure when to receive it, and hopefully the whole timezone magic also works! And yes, the unsubscribe workflow also exists and works :)

Also, what other topics might you be interested in? Happy to hear from you!

Leave your comments on Mastodon, Twitter.

If you want to learn more about my journey and the fediverse, follow me on Twitter or on Mastodon, or follow fedistats.cc account on Twitter or on mastodon.