The Algorithmic Bridge turns two this week.
To think that two years ago ChatGPT wasn’t even a rumor. The world was really quiet back then even if it didn’t feel that way to those of us who were into this already. DALL-E 2 had just launched. Midjourney and Stable Diffusion were weeks away. Crazy to think that the most popular model was still GPT-3.
Those calmer times are gone now and in turn, we pay much more attention to all of this—including my newsletter—for better or worse.
Two years ago, I brought with me 700 brave explorers. We are more than 26,000 today. To some, this won’t seem like a big number (there are people in here with more than 1 million subscribers!) For me, however, it goes beyond what I thought possible. I can only be happy for all of you who have remained loyal readers.
A short anonymous survey
10 quick questions, 3 minutes to complete—and an opportunity to influence the direction of this blog. This survey is important in the sense that I do it once a year. Feel free to jump any questions except the first one!
A couple of changes
The newsletter will remain mostly untouched in terms of style and themes except for two things (I’ll factor in your responses from the survey).
First, I want to focus a bit less on daily, timely news and more on evergreen ideas, essays, and explorations. I want to pay more attention to people, culture, philosophy, and technology itself, and much less to dramas, marketing, and shiny stuff. Some of the latter are key to making sense of the bigger picture but I won’t write about Altman every two articles. Perhaps the best criticism is to not criticize him at all.
Second, I’ve changed my approach to the cover images I use for the articles. I shared my thoughts on Notes:
I’ve decided I won’t be using artificial intelligence to generate my article images from now on.
I’m not sure yet what I’ll do instead. Perhaps nothing, perhaps something like Unsplash, but ideally I’ll commission work Erik Hoel’s style, which I find the most appealing approach (if I can afford it financially, which I don’t know yet).
I believe this move is long overdue. I criticized writers using ChatGPT and found half-baked excuses to not do the same with images, but I was just being a hypocrite scared of facing the reality that I wasn’t doing it right.
I don’t think it’s wrong to use AI for private or personal uses. It’s great. But for public use, especially if I’m making money out of it, it’s not okay. I’ve known this for a long time; just had to make up my mind to be willing to anger the algorithm by not having an image at the top and, more importantly, face my own incoherences.
I don’t do this to implicitly signal that others should do the same. I won’t criticize writers doing this (Substack is a place for writers so it’s not the same using ChatGPT to publish articles than, say, Midjourney to have an image at the top; I’d criticize it on ArtStation or DeviantArt).
I do it to keep things internally consistent. The easy route isn’t always the best. Now that I’ve taken a step back to visualize better what’s happening with this technology I can see my mistake crystal clear and I’m committed to amend it from now on.
It’s fine to be an early adopter but not at all costs. Perhaps in a few years, it will become acceptable if companies offer better partnerships to artists. For now, it isn’t.
Again, I’m not taking a side for what’s morally good or bad here even if you decide to interpret this change as such (which is fine), just a personal desire to not fight an inner conflict when there are so many outside that deserve my attention.
Stats and graphs
I know you love this stuff.
Last year, for the one-year anniversary, I shared a review of my stats:
13,000 subscribers
500 paid subscribers
In one year those numbers have doubled, suggesting a reliable pace of growth.
26,000 subscribers
1140 paid subscribers
Here are the graphs, a visualization of HOW MUCH YOUR SUPPORT KEEPS THIS PROJECT ALIVE. That deserved to be in all caps. Also, for those of you who write, I hope it serves as motivation and inspiration.
You see, the paid subscriber curve grew super fast at the end of 2023 and has now slowed down considerably. I haven’t yet figured out why (or if there’s a reason at all) so the best approach is to try and do work you will find unambiguously worthwhile.
That’s the only North Star I consider worth following: Do the best I can do.
10 misc. insights
You, my dear readers, are spread all over the world, by professions, ages, interests, hobbies, and stances on AI (I believe some of you are staunch skeptics while others are borderline e/acc... with all shades in between, which is really the best I could ask for, variety of opinions to enrich the conversation!)
What this means is that no single lesson I’ve learned during these two years will make sense or be interesting/useful to all of you. That’s why I’ve compiled a list of 10 things that I like to remember myself. It’s for my future reference so consider this a glimpse of a two-year reflection journal entry. It’s about AI, about writing, and about the world.
There’s actually a second North Star I follow when I write: Write about what I love. I find bliss when both stars (the other is “do the best work I can”) align and risk burnout otherwise. When the overlap in the Venn diagram of those two grows, I’m the happiest that I’ve ever been. And I realize you prefer it, too.
After doing this newsletter full-time for two years I’ve realized it’s easy to get lost in a kind of meta-living that emerges from writing too much without doing anything else in a serious, deep-work manner. I think few people are made to write full-time. I’ve set a goal to go out in the world and do work that’s “doing” instead of just “contemplating”.
As AI becomes more popular and more important for world leaders, the incentives that misalign it from the general well-being of society become stronger and thus the possibility to trust any one source of news or any one product or service degrades. Filter things through the lens of covert motivations.
Substack is better than Twitter for now but there’s an inverse correlation between the size and success of a media platform and the internal toxicity that emerges spontanously. It’s an inverse network effect—it gets worse as it gets bigger.
There’s a superpower that comes from reading that I hadn’t realized before I started writing. When I read a good passage or a good novel my ability to write well increases a few times over for a few hours afterward. It’s as if reading awakens some latent power inside me that I can’t seem to access at will.
The amount of people writing about AI is correlated with the probability for the writer of getting some selfish value out of it rather than with the importance of the topic itself. (It was already very important when almost no one wrote about it.)
I’ve realized the need to get external validation from my writing never ends. You may think writers can just work in a vacuum but we all want our work to be read if possible. We want our ideas and our style to travel far and be known.
Making a newsletter work as a job is absolutely not easy. Consistency, good writing, interesting topics, marketing skills, etc. are all valuable but not sufficient. Luck plays a key role—your goal is to minimize its importance by doing all the other things, but don’t forget that it exists to capture the opportunities it creates.
If I had to start all over again I’d do a few things differently. One of them would be not writing about AI (solely). The space is saturated right now, there’s way more offer than demand. The market created the opportunity long ago and has eventually killed it under the weight of its own success.
I will keep doing this for as long as time, money and circumstances allow it.
That’s all for today. I really appreciate you being with me on this journey and hope to see you walk it with me for another year.
The Algorithmic Bridge only exists thanks to your generosity and support.
Congratulations Alberto! Keep up the high signal to noise ratio and authenticity! 🙏
Congrats and done! Keep up the great insights.