Substack, meet Superstack

How AI changes the nature of writing, audience engagement, and making a living as an intellectual

So far, I have been a very content customer of Substack. It genuinely makes life easy as a writer. Audience management and income generation are automated. The drafting interface is clean and straightforward. Censorship is noteworthy by its absence.

Sure, there are a few gripes. The newer social and messaging features feel like bloat to me personally. The interaction between reader and writer experiences can be awkward to manage. It could do more natively to support things like book upsells — a natural need for writers. But those are improvements, not fundamental problems.

After being evicted from Mailchimp, Medium, ConvertKit, GetResponse, ActiveCampaign, and Patreon as ways of reaching my audience, this platform feels like a breath of fresh air. I particularly value that it allows me to continue with my “gift and patron” model of free content rather than paid access.

That said, I’m noticing a growing tension between how I worked from 2003 (when I first started my Telepocalypse telecoms blog) through to around 2024 (when legal work forced me to spend all day, every day, inside a chatbot window). AI is transforming the entire idea pipeline — origination, presentation, consumption, and monetisation.

As a brief thought experiment, I’ve been pondering what a “Superstack” might look like. The problem it would address is how to publish AI-generated content that is valuable and human-readable, but not necessarily constructed in the idiom of traditional essays.

I produce many frameworks, for example. Sometimes I reproduce them here as “executable essays”, but they are closer to products than articles. Many are the result of endless hours of refinement by myself and collaborators. Each becomes the nucleus of its own micro-community of users and developers. It isn’t “AI slop”, but neither is it Sunday newspaper long-form prose.

To bring this to life a little more, here’s how I tend to work now. While I still write occasional traditional essays or do photojournalism, most of my insights arise from stress-testing concepts through AI conversations until I reach an “aha!” moment.

That “aha” might be a short phrase the AI generates — the nub of an issue — even though the machine itself cannot recognise it as golden. Or it might resist an argument and then relent, exposing a deeper assumption worth sharing. Sometimes I notice a pattern in my conversations, surface it to the AI, and it runs with it, helping develop the idea further.

As I go, I often save chunks of AI output into Word documents. I maintain a folder structure that separates projects and litigation from my “AI research lab”, and from the downstream canonised “intellectual infrastructure” of finished outputs — alongside my engagement in civil society. I even use AI to help organise it all into folders, create synthesis documents, and perform adversarial stress tests using other LLMs.

While I still attend occasional physical events and report on them, increasingly my job as a public intellectual involves sharing the creative output of dialogue with a tireless (and sometimes tiresome) synthetic mind.

A “Superstack” might complement Substack, extend its functions, or be a completely different service. It might operate on a different model than the strict publish/subscribe format. Its purpose would be to remove friction from sharing AI-generated content — and monetising it in some form — without forcing it into the mould of essay writing.

It might allow distinct categories for analysis, framework synthesis, and practical application. Payment could be retrospective — “Did you find that useful?” — via tipping. Notifications might be driven by AI-based discovery rather than one-person email feeds.

Perhaps the deeper issue is that Substack is where I speak from the heart. I am the world’s top authority on “being Martin”, so my testimony is authoritative on that narrowest of subjects.

At present, publishing AI-accelerated intellectual content feels alienating to some readers, yet deeply engaging to others. I know people personally who are applying some of my ∆∑ framework tools to holding power to account — sharpening their use of freedom of information requests and auditability pushback, for example.

Right now, there is no easy way for me to add a secondary channel for “things I find intellectually interesting but that may not engage everyone”. I rarely watch videos. I would much rather read an AI summary of a transcript. We all want compression — to fit more into our limited time.

When I write long think pieces, I assume many readers will feed them into AI. If you send me a long email these days, I often let AI produce a structured gist before I read it closely myself. AI is transforming how we engage with written and visual content. The trend will only accelerate. Yet our publishing tools remain locked in a pre-AI paradigm.

Invitation to engage

I’ve left comments open to all. I’m curious how AI is changing how you consume content, run your professional life, or present your work to the world.

My sense is that many of us need a “Superstack” — a way to promote AI-assisted outputs that are worthy of sharing, whether narrowly or broadly distributed.

My concern is that Substack could become like video for me: people send it constantly, and I engage with almost none of it. The risk is that AI overwhelms the system, without a counterbalance to absorb the pressure it places on our attention.

What do you think?