The newsletter now has 20 articles under its belt! We've hit a proof-of-concept milestone that I wanted to get to, where there is enough free content to suggest to new readers what kind of articles they can encounter. There is one remaining subsection I intend to add in the near future; more news when that's ready.

Prev: Newsletter Refurb 2025-08-08

Deep Links on Android

As is ever the case with bootstrapping something new, you make some mistakes and learn some things along the way. Earlier today I cleaned up a bunch of inter-article reference links on the Medium version of the publication. Getting Android deep linking to work for Medium is a little fiddly and I had several articles go out that got it wrong. If you experienced being bounced out of the Medium app to your web browser, my apologies. That should be fixed now. My goal is to make that translation happen via automation so forgetting the manual task doesn’t resurface the issue.

Other Learnings

So far the production process has been almost entirely manual. The AI support has basically been:

• Google Search

• ChatGPT prompting to do searching

• ChatGPT drafting of some of the matplotlib code, which I end up adjusting, particularly if I want PNG files created from it.

Everything else is organically grown. I like the writing, but I have found that at best half my time goes into creating intellectual content, and the rest goes into dealing with the machinery of publishing and making it all work visually. As the material shifts from economics and opinion more towards math and code with supporting visualizations, the split is closer to 25–35% on content for whatever the key concept was about. For example, the animation for the Markov Chains with NetworkX and PyDTMC article involved a 14-layer Draw.io diagram and animated GIF conversion that took about 16 hours before it all worked as I had imagined it, and even then I had to create a static slice for how Substack presents the article on the home page or in a note. Then layer on top of that converting the article to Medium, creating a Substack note, and creating a LinkedIn post announcing the article.

My objectives are to get more frequent content out, improve the quality of visualizations, and market the newsletter better—the last of which is getting very little attention. The solution? Automate more of the process.

The Experiment

For a while I've been a believer that Small Language Models (SLMs) are an under-appreciated resource in the GenAI space. LLMs have their uses, but the industry forces around them have issues that don't necessarily align with better customer outcomes or environmental consequences. I was glad to see NVIDIA release a paper promoting the role of SLMs for Agentic systems. Orchestrated assemblies of SLMs make more sense to me as a way to balance stochastic and deterministic processes with lower power demands and better CapEx/OpEx efficiency.

To that end, I've started an MVP within ChatGPT that I hope to migrate later to a collection of special-purpose SLMs. The structure is an orchestration template with multiple defined roles, where some tasks only happen when I trigger them. Key roles include:

• Mathematician: retrieves formulae and references (LaTeX ready).

• Principal Software Engineer: drafts code based on narrow criteria.

• Technical Visualization & UX Expert: suggests visuals and coordinates implementation.

• Author (me): sole prose source and final decision maker.

• Editor: finds minor text flaws without rewriting.

• Systems Automator: observes workflow and produces summaries; responds to trigger commands (e.g., reference formatting).

Several other roles exist, but these give the gist. This approach is already forcing me to think through organizing responsibilities and interactions among SLMs. As an MVP budget, the only cost is a $20/month ChatGPT Plus subscription.

This article itself is the first, albeit very simple, output from using the template.

• The Editor didn’t get too crazy. ☺

• I realized during the process that I need to give the template a lot more information about constructing visualizations, so that isn’t a factor yet.

• Inter-article linking support isn’t in place yet, that was manual, but the reference retrieval and formatting (below) it handled just fine.

• It doesn’t yet know anything about porting the article to another platform, but we’ll get there.

References

• Belcak, P., Heinrich, G., Diao, S., Fu, Y., Dong, X., Muralidharan, S., Lin, Y. C., & Molchanov, P. (2025). Small language models are the future of agentic AI. arXiv preprint arXiv:2506.02153. https://doi.org/10.48550/arXiv.2506.02153

The Experimentalist : Newsletter Status 2025-09-01 © 2025 by Reid M. Pinchback is licensed under CC BY-SA 4.0

Just a quick note on some long-overdue cleanup now that the newsletter has enough content to guide me on the next steps.

Prev: The Press Room| Next: Newsletter Status 2025–09–01

Why the Changes

New articles are will start rolling out on technical themes, and it was time to help subscribers to find the content that interests them. Performing some cleanup now makes it easier to just follow along with the new pattern as content is released.

What Changed

1. Both the Substack and Medium versions of newsletter now organize material into thematic areas. Those are “sections” in Substack, and “subpages” in Medium. The existence of these is visible both in the home-page menu and in the regions of articles displayed on the home page as you scroll down.

2. The thematic areas are:

   1. Hammer & Anvil: Articles on the tech economy, which already contains several items.

   2. Hot off the Press: Posts like this about newsletter features and usage.

   3. The Laboratory: A new area for home office/small business tech experimentation. Not yet visible, but stay tuned as that’s about to change!

   4. More to follow in time for data engineering, observability, AI and ML.

3. For Substack subscribers this will allow you to control which notifications you get. At the moment nothing is different. You’ll get all of the new article notifications, but you now have the power to exclude notifications for sections you aren’t interested in. In an Attention Economy, I’d just as soon let you enjoy some silence and not sift through unwanted inbox visits. Apologies to the Medium subscribers, but that platform does not have an equivalent feature for subpages.

4. The changes above will only be visually meaningful on a web browser. You likely won’t see a difference on Android app clients for either Substack or Medium. I don’t personally use an iPhone so I have no idea what the outcome is there, but I suspect it is much the same.

5. I tidied up the in-article navigation links on multi-article series to eliminate “View Series”. I already wasn’t using that to link to anything for Medium because Android deep linking doesn’t work in the Medium app for tag page URLs. With the migration to Substack sections apparently the Substack app suffers similar problems for section URLs, so that eliminated the last reason for a “View Series” text fragment to exist.

The Experimentalist : Newsletter Refurb 2025-08-08 © 2025 by Reid M. Pinchback is licensed under CC BY-SA 4.0

Writing online faces challenges, as does the reading of the resulting product.

• Readers have fair claim to a need to understand where an author is coming from, what their personal or professional stake or biases may be, in order to evaluate the integrity and relevance of the content.

• Readers increasingly crave authenticity and less emotional button-pushing than AI-crafted material is often designed for.

• Accuracy and research matter, for which AI can be both a support and hindrance.

• It is a worthwhile community service with negligible competitive risk to help others see what they might face in their own future literary efforts.

Next: Newsletter Refurb 2025-08-08

My Biases

My origin begins with family that for the most part got dirt under their fingernails. Growing up was filled with stories of the life and hard decisions forced by the Great Depression. Most of the extended family and friends were firmly blue-collar, some were white-collar middle class, and some farmers were sprinkled into the mix. Politics were varied and not party-affiliated. Family religious views were all over the map: Anglican, Catholic, Mormon, Salvation Army, Jehovah’s Witnesses, Swedish Presbyterian, and over-the-airwaves televangelist; tossed into that I was the lone Daoist/Buddhist-leaning kid who was into martial arts.

One of the things I value about writing is that I usually sound less energetic than what you might hear when my Scottish, Irish, and English roots combine with growing up around teamsters (which I was myself for summer jobs), carnies, bus drivers, and longshoremen, which then later morphed into many years of the Massachusetts experience. If I get up a good head of steam, I fully admit I can give Bill Barr a run for his money. I try to keep a tight leash on that. Nobody needs another Ted sequel. The delete button really, really helps.

I was one of only a couple of family members that pursued a post-secondary education. The results were… mixed. I’d be on the Dean’s honor roll one year, and trying not to flunk out the next. Somehow in spite of that, I ended up working maybe a total of 18 or 19 years within academia. Initially it was just undergraduate rolls in development (fund raising) and cognitive psych research software development, then post-graduation within distance ed, academic computing support, academic administrative computing, and ultimately an affiliated cancer genomics research group. No, I don’t also do juggling and card tricks on weekends at kid’s birthday parties, but it would be an understandable guess since I just summarized only half of the resume.

All of that got tossed into the blender and resulted in valuing progressive goals as an aspiration for individual freedom and opportunity in our shared human experience, fiscal conservativism so there is some thin hope that the money flow alone doesn’t undermine long-term attainment of those aspirations, and I think a pretty pragmatic expectation that the confluence of education, economics, politics, and religion will always combine to make a common meeting of minds very challenging for society.

In short, I’m probably not really on anybody’s team. Y’all get to be equally skeptical of anything I have to say. I’ve never really got the “team” thing, I prefer to dig into each issue as something that has specific context, its own details or history, and I try to accept whatever conclusions the context and data seem consistent with. I also remind myself I usually get three things wrong every day before I’m even out of bed; given time I hope to cut that down to two.

AI Position

Much like I don’t get excessive team tribalism, I also don’t care about fanboys on really any topic. Setting aside the substance of AI, there is a contingent that only cares that you share their emotional energy. The more I see of it, the more I gravitate away since life is rarely lived in the extremes. My position has, very simply, varied over time. I expect it will continue to do so. Some initial reactions were wrong, some so far are not, and time will continually force me to re-evaluate.

Article Content

I do not have any interest in using AI to generate article content for the simple reason that this is what I enjoy. I cannot imagine a more self-defeating behavior than to hand over the thing I like doing, and instead spending more time on other things I don’t like doing. This is not a criticism of other’s use of AI in writing. It is just a personal choice on where my finite human experience will be devoted. I do not view the act of writing as merely a transactional endeavor. For me the process is just as much the point as the published result, and I’m the one and only caster of that particular vote.

What I do use AI for is generating images to establish the mood of each piece. If I had the budget to fund a fractional graphic artist or cartoonist role, I would much rather that. Time will tell if I get there. For now, I value how AI helps me bootstrap a small refinement in user experience.

Advertising

Some acts of publishing words are not the same. Keeping your content visible means navigating through social media algorithms, and authors have no control over the heuristics those services contain. One of the more pragmatic ways to deal with that is to create shorter posts on various sites which function as advertising, redirecting users to the primary platforms: Substack and Medium in my case. I’m not currently using AI for advertising posts, but it is entirely possible that one day I will. Whether you craft advertisements manually or via AI, they are increasingly a highly artificial exercise and there is little point in pretending otherwise.

Research

There are understandable concerns about the accuracy of research done via AI. On that front, I have good news. I’m a digital packrat. If there was a Digital Packrat Olympics I would have a string of gold medals. I’ve been squirreling away links, notes, projects, and dumps of those into Git repositories for years. I add dozens of tidbits every single day. There are even a few boxes of paper reports tucked away in corners, and don’t even get me started on the books which total approximately a half ton by weight. I know because I’ve moved those boxes.

Given the time it would take to write articles on even a tenth of all that, I’ve got a multi-decade head start before I’m exposed to how online content degrades via AI-driven data recycling. If I have a moat for a competitive advantage in writing, information hoarding may very well turn out to have be it. Enjoy. That said, if I ever did use AI to help me with article content, it could be to digest this immense mound of trivia and help me turn it into starting points for code or topics. Once fully incorporated via LLM or RAG, I’d be back to doing my thing my way.

Wrapping Up

I’ll have other posts about the mechanics of newsletter publishing and how some decisions were made. There is also the potential for others to contribute their own articles and so I’ll be speaking to that process as well.

The Experimentalist : The Press Room © 2025 by Reid M. Pinchback is licensed under CC BY-SA 4.0