hacksciences: Pattern Recognition
notes from my fourth hype cycle
In the early 2000s, I used to write (infrequent) blog posts on my website at hacksciences.com. The site ran on a Dell OptiPlex GX300 (dual Pentium IIIs and 1GB Rambus, natch) under my desk, running Gentoo Linux (probably a 2.4 series kernel…).
Why did I start with that nugget? To let you know it’s been a while. (I mean, that reference was current at the time). I started my career in IT (desktop support and, soon after, Windows NT —> 2000 upgrades), and now people read my resume and don’t even think I’m technical.
The contrast to the era in which I’m starting this blog…er, Substack…is staggering. The levels of abstraction and the thinking power available at our fingertips…I’ll never not be impressed by the degree of technical progress in my lifetime.
I’m writing these entries on an M4 iPad Pro. This first one, I am writing on an airplane with actual good WiFi. To think: I went to college in 1994 — the public web was pretty much brand-new; started out in “IT” in 1999; my studio apartment got my neighborhood’s first DSL connection (Pacific Bell had to send three guys out); had to switch off of Mac Classic to Windows NT and Linux in order to do real work; ran back to Mac after OS X came out; witnessed the invention and popularization of smartphones, Web 2.0, the cloud, AWS, SaaS and PaaS, social media, and now the generative AI revolution. I use a laptop daily that has 64GB of RAM; my first computer had 64K.
All that to say, I’ve developed some pattern recognition.
On this Substack, I’ll share some thoughts on where we are, plus the occasional lesson I’ve stolen from very smart bosses. I consider myself really lucky, but man, I’d love to have known some of this stuff a little earlier.
There are great tech writers out there, and great business writers — I’m not trying to be them. I hope I can make my unique perspective - an IT guy who grew up through just about every functional role imaginable to become a “P&L guy” and enterprise leader - interesting enough to read once in a while.
You might be familiar with the Gartner Hype Cycle.
Right now we are passing the very Peak of Inflated Expectations of the gen AI hype boom. It’s a little harder to recognize this time as “Inflated Expectations” — large majorities of Americans consistently say they hate and fear the technology, as companies laying off 25% of staff blame “AI” (rather than management profligacy), and Dario Amodei and Sam Altman stoke fears (of job losses and of China) in order to raise trillions in capital.
Why is it hard to see? Because, really, very few businesses have deeply adopted the technology in productive ways. And those who have - many of them are swimming in internal AI slop, created by corporate creatures who simply want to get recognized for “using AI” and simply churning out more work, more prototypes, more vibes. When the credits get used up, and the value hasn’t been created, the slide into the Trough of Disillusionment will begin.
It’s also hard to see because this is all happening against a very challenging geopolitical backdrop. And, whatever your views on policy, the domestic political environments in the US and China are trending in a direction as autocratic as they’ve been in some time. Conflict over Taiwan could fundamentally alter the AI trajectory in ways I don’t think we’re talking about very much.
From past hype cycles, I can tell you the most interesting outcomes won’t be visible to most of us until well into the Slope of Enlightenment. The defining rivalry of the Rise and Peak phases, OpenAI vs. Anthropic, will be a fun memory in half a decade, I think. And losers of one phase can always become winners of a later one (see: Microsoft, Apple).
Thank you for reading. And I hope you’ll watch this space.