Scales and surprises

I attended the conference known as Experimentation Island last week and it was exactly what I needed. Fantastic people, fascinating conversations, and fond memories. I'm truly lucky to have a professional community filled with so many eminently decent people; it may have only been a few days, but it was time together that felt terribly important.

At some point in the last decade I erased the line between skepticism and cynicism. I've always tended toward believe-it-when-I-see-it and it's largely served me well. Most of my managers mark my caution, consideration, and deliberation as strengths and I like to think it's how I've been able to build so many trusting relationships with colleagues, stakeholders, etc. But after I got back and shared how uplifting the conference was, a coworker make a joke in chat:

What does joyful Matt even look like?

I knowingly lean into the cantankerous and neurotic because it can be fun, but it stung. Do I actually want to be the bearer of bitterness or herald of harshness when I show up?

Probably not. But I have been, for a while. And that sucks.

That being said, why shouldn't I be a bit jaded? The last decade or so hasn't had any shortage of things to be bummed out about. Trump II has more of less been a speedrun of everything we doomsayers doom-sayed. A lot of us are still feeling the hell of the COVID years. The pervasive sense of polycrisis is more than plausible; it can feel downright reasonable.

And in being so reasonable is the riskiness. It can feel downright cozy to retreat into your priors. I've certainly done so. I've found a lot of reassurance in moving my lurking to Bluesky, where there are plenty of people struggling in the ways I think I am. But we're not beating the charges that many—if not most—are posting their way through their anxiety, depression, and more. It's a hell of a feedback loop. Suffering alone sounds like hell, but when schedule and circumstance force some offline time... it's usually not nearly as bad as I'd thought.

So when I'm confronted with a room of people I admire and cherish after spending most of my time working from home, alone, for years, some stuff starts to click.

The frothy hype waves of the late twenty-teens—crypto, blockchain, metaverse—never passed the sniff test and I think I got it right on most counts. And as I watched the tech industry wrap itself around AI I assumed it was pretty much the same story: the smartphone and the app are starting to sputter, so where do we drill the next well? (And who do we pass it off to if we haven't struck big?) However, several things can be true:

1. AI is part of an unprecedented concentration of wealth, power, and technology that so happens to be midwifing a global resurgence in fascism.
2. The current crop of AI tools are new, bizzare, and exciting human-computer interfaces.
3. Nobody can know if it will ultimately be good, bad, or something else altogether.

And that last point is pretty important, because I don't think it's productive to equivocate about the known harms. (To name a few: social atomization, sycophancy, and psychosis; environmental impact and carbon footprint; economic disruption and disempowerment.) But this bell is not going to be un-rung. And, in the spirit of skepticism-over-cynicism, I've been trying to pry my lids back open and Paul Ford's essay from last fall certainly helped:

I love normal technologies. They come with manuals. They change periodically, but you can build craft and professional skills around them. Bubble technologies change constantly, and there is always a threat that they will either destroy society (bad) or make everyone besides you wealthy (worse).

I disagree with the ranking—there's nothing I enjoy more than a little abjuration of the self, after all—but I'm dying for the novelty to burn off. I think the VCs bleating that will "democratize talent" are full of shit, and the cheap, noisome results speak for themselves. (If you've sent me generative photo edits I've tried not to hold it against you, but I'm not finding it easy.) This thread spoke well to the dynamics of vibe coding and software design, and it generalizes well to a lot of current AI use-cases.

Leaping to high-fidelity across multiple fronts is flashy but it doesn't mean you've actually solved for anything. I know this is classic "outcomes over output", but the frequency with which correspondents are surprised by the premise of AI-assisteed processes being noisy expansion/lossy compression cycles has been easy prey for the monstrous cynic within me.

As I'm usually a bit of a hermit I picked up a case of the conference crud last week and have slowly been on the mend. As I've been working through the fog of fever I've also been progressively probing Claude to see if I might figure out exactly what's been bugging me. And I think it finally clicked: I'm terrified I might cheat myself out of learning something.

Despite my chagrin I actually do think I enjoy being proven wrong because it means I've learned something I didn't know before. At the conference I was lucky to be on a panel discussing clarity and complexity in experimentation last week, and as the conversation progressed I found myself drifting toward a conclusion: answering the questions front of mind might be a less interesting exercise that identifying the assumptions I've made without realizing. And AI tooling provides an immediate release valve taht directs us to the former rather than the latter.

I might not look it but I am a cyclist but I am, and this AI stuff might be a bit like how I feel about e-bikes. By riding one you get a distillation of the freedom, fresh air, and flexibility of the bicycle. But in doing so, you're losing something: the strain, rush, and rawness of being embodied. To feel limits is just as important as the freedom. And as amusing as it is to see Copilot, a Microsoft product, determining the path of least resistance to creating a PowerPoint deck is to write a wildly complex Python script while failing to capture the actual outline I provided it, the pleasure is fleeting. It's not riding—it's commuting.

I've been working through how to rebuild this site yet again because I can't leave well enough alone, and I've found it very satisfying to prompt Claude with a one-two punch: here's what I'm currently thinking, help me find the solutions; what are the questions I should be asking but haven't asked so far? It's definitely not a perfect solution and I'm still a novice at all this, but it feels far more satisfying and I'm excited to carry that spirit forward.

It's been a long, long time since I've taken the time to do this sort of thing. Writing is thinking, and thinking should take work. To struggle toward expression is maybe the most human thing we can do. And it feels more important than it's ever been to find what is human and to cherish it. And it's nice to give myself permission to get a little excited to ride again.