Editorial Hello:
Today’s note is a reader update to explain why my usual aggregation email has basically gone AWOL the past few months. It’s not that I haven’t been compiling it. I have. I’ve just found it increasingly difficult to turn it around quickly enough to keep it relevant. There are what feels like a thousand unfinished drafts in the back-end of The Peg, that never quite made it to publication.
It reminds me of the Sphinx Gate in The Neverending Story. In the film, the sphinxes judge whether someone is worthy of passing. My drafts are like the knights who try to pass the gate but continuously fail to make it.
The film never explicitly reveals what makes someone worthy.
But as I commented on X the other day, “the problem of our time is not merely information overload. It is that the time required to construct a coherent understanding of reality is becoming longer than the time during which that understanding remains valid.”
The production of this email was one of the factors that inspired that comment!
On average, it was taking me a good 3 hours (depending on the day) to find all the links, read all the stories, absorb all the reports, and then another 2-3 hours or so to put them all together and structure original thoughts around them. In recent weeks — I suspect due to the growth of AI-assisted output — that process has only been extending.
That’s why I have finally bitten the bullet and resorted to throwing AI at the problem myself. I have spent the past few weeks vibe coding my own AI-assisted aggregation mechanism. The idea is that it will do most of the heavy lifting with collation and aggregation. I will then add value by including my original thoughts where relevant and/or links that passed the system by, identified through old-school journalistic processes.
This should, in theory, free me up to do more original content, without limiting my exposure to news and developments — vital to maintaining an expertise — since I will still be editing the AI output.
I hesitated to go down this path before because I worried subscribers would not see value in an AI-aggregated newsletter. But I have decided that AI-assisted output is better than none (the alternative increasingly was writer-overwhelm-induced paralysis)!
The value proposition, as I see it, is as follows. While any of you could, in theory, go through the same vibe-coding process to produce an equivalent newsletter … configuring it to the same standard would take a significant time investment. Moreover, catering to unknowns unknowns versus known unknowns is not easy to control for. A good overview requires more than asking the AI to scrape the internet for stablecoin news. It requires expert insight and guidance on where to look, and access to private correspondence from sources and contacts. That’s journalism. My AI-assisted newsletter will thus blend all these expert-guided inputs together. It will then add even more value by having me refine and edit the final output, while adding my own insight where it is relevant.
The economics should also make sense. Generating a bespoke AI-generated newsletter of this complexity costs tokens. Network effects and division of labour, however, create an arbitrage opportunity. In theory, it should be far cheaper for readers to access the newsletter at my subscription fee (which also gives them access to all the other original content I produce) than try to restructure it on a bespoke basis themselves.
So that’s what to expect. Until then, here — for my own peace of mind and completeness — is the final amalgamation of all drafts that never quite made it the past week:
