45+ Years in Technology: A Journey

I still have those books. Dog-eared, coffee-stained, held together by stubbornness and nostalgia. They're from the late 1970s - programming manuals for BASIC, Pascal, and C. I was just a kid when I first cracked them open. Getting my first computer a few years later meant I could finally run the programs I'd been tracing with my finger.

I didn't understand everything. Hell, I didn't understand most of it. But something clicked. While other kids were playing outside, I was hunched over a keyboard, trying to make a computer do what I wanted. Not just use programs - I wanted to create them.

That obsession never left.

Updated January 2026: Added pendulum physics framework and Monday Morning Checklist.

The Pendulum Physics

History does not move in a line. It oscillates between centralization and decentralization. The pendulum always swings back.

• 1970s: Mainframes (Centralized). IBM owned the world.
• 1990s: PCs (Decentralized). Power moved to the desktop.
• 2010s: Cloud (Centralized). Power moved to AWS/Azure/GCP.
• 2030s: Edge/Local AI (Decentralized). Power moves back to devices.

We are at peak centralization right now. The cloud providers have more control over computing than IBM ever did. But the physics demand a swing back. Local LLMs, edge computing, on-device AI—these are not trends. They are gravity.

If you are betting on "more cloud," you are betting against the pendulum. I have watched three complete cycles. The fourth is already starting. The winners of the next decade will be building for decentralization while everyone else is still optimizing for cloud.

The Machines That Made Me

The computers of that era were brutal teachers. We're talking kilobytes of memory - not gigabytes, not megabytes, kilobytes. When the TRS-80 launched in 1977, it came with 4 KB of memory and cost $599. Processors so slow that today's smartwatch would embarrass them. Every byte mattered. Every CPU cycle was precious.

This wasn't a limitation; it was an education. When you only have 64K of RAM, you learn to write tight code. You learn to optimize. You learn to think before you type. Modern developers spin up a container without a second thought. I grew up counting bytes like a miser counts coins.

That discipline stayed with me. Even now, with virtually unlimited cloud resources, I still write code like memory costs a dollar per byte. Early computing taught me that elegant code isn't about what you add - it's about what you can remove. Old habits die hard. Good habits shouldn't die at all.

The Underground: BBS Culture in the 1980s

Before the internet went mainstream, there was something else. Something weirder, more chaotic, more alive. Bulletin Board Systems - BBSs - were the original social networks, run by hobbyists out of their bedrooms on donated phone lines.

I wasn't just a user. I was a SysOp - system operator - running multiple boards throughout the 1980s. My boards were known for the latest door games and solid FidoNet connectivity. Door games were third-party applications that ran on BBSs, and yes, mine were "heavily modified."

FidoNet was magic. A store-and-forward messaging system that let BBSs exchange messages overnight via scheduled phone calls. It was email before email. Social networking before Zuckerberg was born. It taught me more about distributed systems and community management than any computer science course could. Combined with what the Navy taught me about perspective, these early experiences shaped how I approach systems today.

1993: The Internet Changes Everything

I've been on the internet since 1993. Not the World Wide Web we use today - the raw, weird, text-heavy internet of Gopher, Archie, and Veronica. I learned the protocols by necessity: HTTP, FTP, NNTP, SMTP, IRC, Telnet. I didn't just use these technologies. I was building with them.

Over the years, I've set up and maintained:

• Web servers, FTP servers, DNS servers, mail servers, IRC servers
• Several hundred domains —when .com registrations were actually hard to get)
• Custom web crawlers - blindingly fast ones that could index millions of pages
• NNTP crawlers for Usenet archival
• Raw socket applications that talked directly to the wire

I've written my own web servers from scratch. Not because I had to - because I wanted to understand exactly how HTTP worked, byte by byte.

The Home Lab That Ate My House

By the mid-to-late 1990s, things had gotten a little out of hand.

At home, I maintained a network of over sixty computers. Not a typo. Sixty machines, mostly servers, all connected to two full T1 lines. Not fractional T1s - the real deal. This was my personal laboratory, my testing ground, my obsession made manifest.

My electric bill was criminal. My neighbors thought I was running some kind of operation. (They weren't entirely wrong.) But that home lab taught me more about scaling and redundancy than any enterprise job ever did. When you're responsible for sixty machines on your own dime, you learn to automate or die.

Enterprise, Mainframes, and the Stuff Nobody Wants to Touch

Not all technology is glamorous. I've spent significant time in the trenches with systems that make many developers uncomfortable.

Microsoft IIS since version 1.0. Server Side Objects. Active Server Pages when ASP was cutting-edge. ISAPI DLLs and extensions. Microsoft Index Server, Transaction Server, Commerce Server - the whole enterprise stack before .NET existed.

But I'm not a Microsoft partisan. I've done hard time with Apache, NGINX, Hiawatha, and various other web servers. I've written CGI scripts in batch files, Perl, Python, C, PHP, and - God help me - Visual Basic.

And then there are the mainframes.

AS/400 (later renamed System i). ES/9000 series. FORTRAN. RPG. COBOL. These aren't technologies I brag about at parties. But when a client has a legacy system running since before I was born? When they need someone who can actually work with it? I can work with it.

The Cloud Era: 3,000+ Instances and Counting

In 2014, at ZettaZing, I found myself managing over 3,000 Amazon EC2 instances.

The scale had changed. The principles hadn't. Efficiency still mattered - at 3,000 instances, a 1% optimization meant real money. Reliability still mattered - distributed systems fail in creative ways. I'd been preparing for this since my sixty-machine home network. Automation wasn't optional - it was survival.

The cloud didn't make operations easier. It made operations possible at scale. That's a different thing entirely.

2015: Burn the Ships

In early 2015, I made a decision that most people thought was insane.

I sold or donated everything I owned in the United States and started traveling the world while consulting. No apartment. No car. No stuff. Just a laptop and a carry-on bag.

I lived in Bangkok for over a year. Traveled across Southeast Asia and Australia. Worked from co-working spaces, hotel lobbies, and beachside cafes with questionable WiFi. The "roaming" in RoamingPigs isn't branding - it's biography.

In late 2017, I became Chief Architect of SmartEar, Inc while continuing to travel. By 2018, I was exploring Mexico, Central America, and South America. I stayed closer to US time zones while maintaining the nomadic lifestyle.

Never Stop Learning

Here's the thing about technology: if you stop learning, you're dead. Not metaphorically dead - professionally dead.

I've made it a point to stay current through every major shift:

• The Altair and the birth of personal computing - I was there (1977 was the pivotal year)
• The internet explosion - built on it from day one
• Cloud computing - scaled to thousands of instances
• Mobile revolution - adapted and shipped
• Blockchain and cryptocurrency - understood the fundamentals, built on the technology
• Machine learning and neural networks - from academic curiosity to production systems
• ASR (Automatic Speech Recognition) - Chief Architect at SmartEar
• LLMs and generative AI - currently building with Claude, GPT, and open-source models

Every few years, someone declares that "everything has changed" and the old guard is obsolete. They're always half right. The tools change. The platforms change. The hype cycles come and go. But the fundamentals of good engineering? Those haven't changed since I was counting bytes on a machine with 64K of RAM.

Now I can apply those fundamentals to whatever the current "hot" technology happens to be. I've been doing exactly that for four and a half decades.