Writing

I’ll be honest: when my Flipper Zero arrived, I opened the box, turned it on, pet the dolphin a few times, and then had absolutely no idea what to do next.

This is a $170 device that can read NFC tags, capture radio signals, clone RFID badges, act as a universal TV remote, and inject keystrokes into a computer via USB. It’s genuinely one of the coolest pieces of hardware I’ve ever held. If you haven’t seen one, it looks like a chunky orange Tamagotchi from 2003 with a tiny monochrome screen and a digital dolphin that levels up the more you use it.

For the first two days, it mostly sat on my desk while I scrolled Reddit threads titled things like “what should I actually do with this thing?”

Here’s the problem with the Flipper Zero: the distance between “this can do incredible things” and “I know how to make it do those things” is vast. The official docs are decent. The community is helpful. But you’re jumping between radio protocols, modulation types, DuckyScript syntax, and what feels like hundreds of community firmware options all at once. The learning curve isn’t linear. It’s more like a pile.

I learn best by building, and I learn second-best by having things broken into small pieces I can work through at my own pace. So instead of trying to absorb everything from scattered forum posts and YouTube videos, I built myself a curriculum.

I wrote a while back about my complicated relationship with technology. How I love it, feel crushed by it, and I’m trying to carve out spaces that feel good. I ended that post by saying I hadn’t gotten into the other layer yet. The one where it’s not just about attention and noise, but about ethics.

This is that post. Every platform comes with an asterisk now.

I was on Substack for a while. It’s a good product. Easy to use, nice reading experience, built-in discovery. I moved my newsletter there because the friction was low and the audience potential was real.

Then in November 2023, Jonathan Katz published a piece in The Atlantic reporting that he’d found at least 16 newsletters on Substack using overt Nazi symbols in their branding, with dozens more promoting white nationalist ideology. Some of them had thousands of subscribers. Some of them were monetized, meaning Substack was taking a 10% cut.

What followed was a slow-motion reckoning. 247 Substack writers signed an open letter asking the company to explain why it was platforming and profiting from this content. Substack co-founder Hamish McKenzie responded by arguing that removing Nazi publications would amount to censorship and would make extremism worse. Casey Newton at Platformer announced he was leaving the platform, taking one of Substack’s largest newsletters with him.

It’s 11:47 AM. I’ve already triaged Slack, answered a pile of emails, sat through a standup, and made way too many small decisions before lunch. My brain is cooked.

And now someone’s asking: “What do you want for lunch?”

I don’t know. I genuinely don’t know. Not because I don’t have preferences, but because the part of my brain responsible for preferences has clocked out. It gave everything it had to “should we use the existing API or build a new endpoint” and now it has nothing left for burritos vs. sandwiches.

So I built a thing.

Decision Maker is aggressively simple. You type in options. You click a button. It picks one. That’s it.

I built a French Press Calculator a while back because doing ratio math before caffeine felt like a problem worth solving. It works. It does exactly what I wanted it to do. And I use it maybe twice a week (generously) because most mornings I just eyeball it or have already set the drip the night before.

Which is the whole thing, really. Building the calculator and actually making it part of my routine are two completely different projects. The first one was fun. The second one is boring and ongoing and arguably the only one that matters. I only finished the first one.

I do this constantly. I have a folder of project ideas that is embarrassingly longer than my folder of shipped projects, and within the shipped folder, “things I actually use consistently” is a much shorter list. I know this about myself. I still open new tabs.

I’ve watched this exact pattern play out at every company I’ve worked at. The exciting part is always the evaluation, the implementation, the launch. Milestones happen. Slides get made. Someone gets credit. Then six months later the data’s a mess, half the team found workarounds, and leadership is asking whether we should evaluate a new tool. To solve a problem the current one handles fine, if anyone had spent time maintaining it instead of chasing the next shiny build.

I wrote about discovering a rogue email platform at a previous company. The team that built it wasn’t doing anything wrong. They needed a tool, found one, and made it work. The problem wasn’t the building. The problem was that nobody maintained it. Nobody documented it. Nobody audited whether it still made sense three years later. By the time I found it, the shared login had been passed through four people and the segment logic lived entirely in one person’s head.

At a previous org, I discovered a team had been using a standalone email platform to send communications to customers. Not the approved MAP. Not something IT had vetted. A completely separate tool, a relic of a long ago acquisition, with a single shared login, manually imported lists, and years of accumulated segments that existed nowhere else.

My first instinct was simple enough: migrate them to our actual platform. Consolidate the data. Get everyone on the same system. Easy.

It was not easy.

When you crack open one of these rogue platforms, it tells a story. Not a clean one. More like walking into someone’s garage workshop where everything is covered in sawdust but they swear they know exactly where the 10mm wrench is.

The list names alone were a journey. There were at least a few variations of “List Name_DO NOT USE” which, if you’ve been in this world long enough, you know means someone sent to the wrong list at least once. My personal favorite was a particular segment of customers that was apparently on its third iteration, helpfully identified with “_V2” at the end.

Growing up, I was the one in my family who wanted to know how things worked. Not just use them, but understand them. I built my first computer as a pre-teen and still remember the rush of watching it POST for the first time. I also remember when it caught on fire because I’d routed a wire too close to the GPU, but that’s a different story.

That curiosity shaped everything. I wanted to know how websites were made, so I learned. I wanted to understand how applications worked, so I tinkered. Technology wasn’t just a tool. It was a door to something bigger. The internet, especially, felt like an infinite library staffed entirely by weirdos who were happy to teach you things if you just showed up and asked.

I spent countless nights falling down rabbit holes. Learning about topics I didn’t know existed. Finding communities of people who cared deeply about obscure things. Playing games that let me wander through worlds someone else had imagined into existence. It was formative in a way that’s hard to overstate. The internet made me who I am.

And now I have kids. And I don’t know what to do with any of this.

It’s not that I think technology is bad now. That’s too simple, and honestly, kind of boring as a take. I still love building things. I’m writing this on a website I built myself. I have a 3D printer in my garage that is running constantly. I get genuine joy from making a small app that does one useful thing. Or a bunch of useless ones. The curiosity never left.

I’ve been evaluating, implementing, and occasionally ripping out marketing technology for over a decade now. CrowdStrike, Teradata, Coursera, and now Accruent. Different industries, different scales, same patterns.

After enough vendor calls, demos, and post-mortems, you start to notice things. The sales team’s job is to close the deal, not to make sure you succeed after. Here are the red flags I watch for now.

Translation: you’ll be writing custom code forever.

A flexible API isn’t a feature. It’s an admission that the product doesn’t do what you need out of the box. Every “just hit the API” answer is future engineering work that doesn’t show up in the license cost.

Ask instead: What can I do without the API? What do most customers accomplish in the UI alone?

You know that feeling. You’re three hours into designing a solution. There are diagrams now. Maybe a spreadsheet mapping dependencies. You’ve considered edge cases that have a 0.3% chance of occurring. The voice in the back of your head is whispering something, but you’re not listening because you’re too busy adding another layer of abstraction.

The voice in the back of your head is saying: this is too much.

You ignore it. You keep building.

I’ve been there. Many times. So I built a quiz to externalize that voice.

Am I Overengineering This? is exactly what it sounds like. A few honest questions. An answer you probably already know but need externalized.

So the world is a lot right now. You’ve noticed. I’ve noticed. We’re all just… yeah.

Anyway, I made a website where you zoom infinitely into a potato. There’s another potato inside. And another. It’s potatoes all the way down. I don’t know what to tell you. You can also receive a random compliment from a very specific person. There is a Victorian ghost out there that really believes in you.

Weird Little Joy Machines is a collection of tiny, pointless joy machines. The whole point is there’s no point. You click around, maybe smile, close the tab, go back to your life. That’s it.

I wanted it to feel like the internet used to feel. Just weird stuff someone made because they wanted to. None of it will help you finish that thing at work. But that’s the whole point.

A few years ago, I was knee-deep in a marketing tech stack migration when someone from IT pinged me: “Hey, did you update the SPF record for that new tool?”

I… had not.

Cue the frantic DNS lookup, the realization that we were one include: statement away from hitting the 10-lookup limit, and the growing suspicion that nobody actually knew if our DMARC was configured correctly. We patched it together, moved on, and I made a mental note: someone should build a simple tool for this.

That someone turned out to be me. Several years later. I started this project ages ago, got 60% of the way there, and abandoned it like every other side project in my graveyard of half-finished ideas.

SendReady is a free email deliverability checker I built to validate the stuff that breaks email delivery: SPF, DKIM, DMARC records, and the one-click unsubscribe headers that Gmail and Yahoo now require.

I’ll be the first to admit that doing quick calculations in my head has never been my strong suit.

So when I got my new french press (an Espro P3, if you’re curious) and the instructions threw out ratio suggestions based on roast level, I got a headache. I just wanted coffee. Instead I was staring at a bag of beans trying to do division.

So I built a calculator to do the math for me. No thinking required, which is the whole point before caffeine kicks in.

You pick your roast level, which sets a default ratio. Light roasts are denser and extract differently than dark roasts, so they need less water. A 1:15 ratio versus 1:17 for dark. I knew the basics, but building this sent me down a small rabbit hole confirming the details.