Hi.

Brainjolt is one of those once-in-a-lifetime things. The only metric of success during our Organic Era was “did this go viral” — a video under a million views was a failure. We created tens of thousands of posts over 13 years. I want to come back to this and put some of our more viral projects, but here’s a couple I remember offhand.

6 million people saw our post about toast in 2016 (it would grow to 300,000 comments):

There were days we’d have 50,000 people real-time on the site all-day long (this was very stressful with just one $200/mo server):

In 2013 The Atlantic ranked us as the 2nd most shared website on Facebook:

ChatGPT launched and GPT-3.5 made coding using AI pseudo-possible. Torn between turning around a failing business or playing with AI, I chose AI (duh!) and began pushing its limits in a series of Tiktok videos that ended up going viral with over 5m views. Some examples:

• AI codes an analog clock
• AI codes simulated raindrops
• Code art with ChatGPT
• AI Codes a starfield simulation

The next month, alas, I had to choose the failing business and so couldn’t play with AI as much for a while.

I uploaded a video of my cat Hobbes to Youtube in 2011. The video was originally from 2007 but I laughed at it enough that I thought others might enjoy it, and it was a time the internet was primarily made up of cat pictures.

It went viral-ish on Youtube itself (800k views) but then it got picked up by America’s Funniest Videos (and Pets) so it got millions of views on TV. They paid me like $250 or something and I thought that was cool and funny.

This wasn’t my first brush with virality — but previous times had been through pseudonyms or entities — and even this time was really Hobbes going viral, not me … but it still scratches the itch, but that only makes it more itchy. That itch would end up growing into an obsession with viral sites, culminating with the creation of Brainjolt with Abraham in 2013.

(placeholder)

Just a lil website I made without using a database and only GitHub Pages + Actions

A simple visual representation of how much year is left. Updates every minute, so it will auto-increment. Mostly just an experiment to play with the latest AI coding tools (in this case, Cursor + Claude 4 Sonnet).

• Version 1 (Sep 2024)

My first Claude artifact. I wanted to see how using Claude worked compared to my normal workflow (as of this writing, Cursor using GPT-4o). Claude first wrote it in React by default, and I know that’s the standard these days, but I really don’t like it and Claude couldn’t get it to work right either. Inputs kept failing, words wouldn’t shuffle right, modules wouldn’t load… it was just confused.

I told it to convert it to jQuery (modernism be damned) and that made it a lot easier for Claude AND me. It worked better after the conversion, no other changes, proving once and for all that jQuery > React. I cleaned up some bugs manually and added some new features (emojis on screen for score, removing them when wrong bc everyone needs consequences, etc) and then fed it back into Claude and published it.

Before this she didn’t know how to type on a computer but after a few rounds she got fast! Hopefully this will help with sight words, too.

It’s incredible how you can make “throwaway apps” for any use. Maybe a better term is single-use app? Or in this case, game. To find Zizi a typing tutor at her level would have taken even longer and the game itself would have been way too overstim — sound effects and explosions and ads and whatever — I needed this app for a moment in the mountains where Z expressed interest in typing, and she was way more engaged since it said her name and were all words that meant something to her.

In 10 minutes I had an app she could play with and that was so thrilling for her, plus she watched me make it so she’s learning it’s possible to build things herself.

My second project using Hatch. They’re building a super cool platform. I’ve spoken to a few of the folks who work there and am really impressed with the caliber and dedication and passion. I want to work with more people like this!

One of their devs even sent a 9 page document back to me on how I could make things more efficient using their features.

At first I tried using Tone.js for this, but that proved too time consuming/confusing so I did what I always do when something is taking too long: hack it.

I used a technique that changed the playback speed of the piano note (or the meow when in MEOW MODE) in such a way that changed the pitch to be different notes.

I got the core method from “Pitch shifting in Web Audio API” by Tuomas Siipola which I encountered in my research and thought was very clever.

“I’m only clicking cause I like you and support you” —Haley

Then I got sidetracked for a bunch of reasons I hope to go into someday but even that is just an excuse, it took me like two hours to upload it to github and make sure it all worked etc. Everyone has two hours a day if they just delete ((insert current social media soma of choice)).

It’s probably not ACTUALLY the world’s first, but I couldn’t find anyone that had done a full letter set using gen AI, or maybe there’s so much ai slop in the world now that things are getting hard to find in 10 second google searches.

The font here is actually individual image files, hand cut like french fries, but using Photoshop and a weekend.

I can’t remember why I even made this.

A popup game / rickroll. Based on the windows dialog game I made in Dec with new features:

1. A start button
2. IMAGES!!!
3. Dialogs autosize dynamically based on image size
4. MUSIC!!!
5. Faster touch response
6. MOAR RICK!!!
7. Dialogs close based on the X not a button
8. Dialogs can be moved in front by clicking on them (z-index magic)

There are still some edge cases of bugs but I am too lazy to fix them and want to move on to the next useless website. If you find one, fix it and send me the code ¬‿¬

I discovered BeepBox a while back and kept wanting to play with it — a simple html/css/js tracker is just the thing to cheer anyone up. I grew up in awe of Screamtracker listening to Purple Motion who was a member of Future Crew. FC created Second Reality (youtube) in 1993 which I grew up watching on my Gateway 386DX/33 over and over because IT WAS SO COOL and imo is still the best demoscene thing ever created. Anyway back to BeepBox… I started at 6am with the idea of just playing with it for a few minutes and 6 hours later this silly homage to American Football’s “For Sure” was BeepBox’d out. (Bonus: An incredible deepdive into the source code of Second Reality by Fabien Sanglard.)

I think this is the first original game I’ve ever shipped (?). I’ve made dumb little things for sport but I think this is the first that actually qualifies as possible a real game, tho even that is a stretch. My high score so far is 1790… can you beat it?

Inspired by Marvin Minsky's Useless Box, if you leave this running long enough on auto mode it will eventually overload itself and thus turn itself off. This is its only purpose.

Over the course of two months me and ChatGPT created a full-featured website discovery platform using React and Firebase. I hadn’t done true head-to-keyboard programming in years and it was a perfect challenge to learn modern dev tools. It launched in private beta with a full suite of features from profile creation, posting, search, tagging, and commenting.

Inspired by Chuck Close’s portrait painting, I made an app that lets me generate hundreds of images in order to make one image. It’s a rather pricey and convoluted way to do this, but it’s also really, really cool. Here’s an example (each tile can be tweaked / regenerated / restyled):

The interface isn’t as fun as I’d like, but it serves its purpose:

when you click a tile, you can choose between previous generations or make new ones:

as with most of my stuff, it runs 100% in the browser

this was by far the most advanced app I had made by myself since my actual coding days, and it gave me the experience to make my next one — The BAD.

While in Copenhagen, I went to the Arken museum with Abraham. There was a wall of paintings and photographs from Frederik Næblerød. I was really taken with it (the concept, not really the art itself, but maybe the whole WAS the art?) and decided to do something similar on my office wall with stuff from books or pictures I had lying around.

What I thought would be an afternoon turned into a weekend turned into weeks into months.

I started taking a timelapse made super hacky by just scrolling through some photos on my phone:

I finally finished it, in a sort of way that a painting is done when you give up on it, and I’ll try to remember to post a final timelapse here.

Inspired by this guy in the Time Bomb Y2K documentary, Emily and I made our own hats celebrating the survival of the human species from the Y2K bug. Get your own at the Hypertext Foundation shop!

My first children’s book, designed for one special kid. It was surprisingly hard to find a good way to make something that works on all platforms (desktop, tablet, and mobile) and plays well with tiny fingers.

After discarding custom website options and storybook creator apps I tested an html export of Keynote which worked surprisingly well. But I wanted to publish changes easier, so ended up using Google Slides using a public link.

I really loved how it turned out. I used a mix of DALL-E 3 and Midjourney for the images, then extended and fixed them up in Photoshop.

I was nervous to show her, but she loved it! Read it a bunch of times, interacted with it (the zoom in is a nice feature), and then talked about it into the next day. (Update weeks later: she keeps asking for it! “Let’s play the Princess Game,” she implores.)

Read about it more in my newsletter #16.

Palm Springs w/ Emily and Abraham: met a father and son who told us they take pictures of the dad jumping and the kid on the floor (or was it the reverse?) and we were so taken with it we started the ritual ourselves.

The first one was in Palm Springs in 2022:

and by the pool:

2023 in LA:

2023 with Hobbes:

2025 in Seattle:

2025 in Hawaii:

In partnership with Hypertext Foundation. Emily has this print in her living room and reportedly she gets a lot of comments about it.

This one never made it to an actual ad because it might have broke Facebooks TOS since drowning could be considered an act of harm, but if you’re drowning in deals, isn’t it a good thing? I didn’t want to entrust this difficult dillema to the Facebook bots, which turn things off just for the fucks of it.

As soon as I got early access to DALL-E 3 I stayed up late for weeks and created thousands of ads, stress testing their system and my sanity. A few of my favorites are on the Instagram pop-up gallery.

Later we compiled the best ads into a book. We printed a bunch of test books and I’m being too much of a perfectionist cause there’s one more tweak I want to make… by the time I’m done it won’t even matter, we’ll be drowning in the stuff. But for its time, it was super cutting edge and almost no one was making ads using this or talking about it, so that gave us a solid year before others started catching on.

Internal tool for Brainjolt

A sequel … or twin? … to the BAM.

It started out as the Brainjolt Ad Desquareifier (BAD!) because we needed thousands of square ads redesigned as vertical ads.

But then more features got added so it became the Brainjolt Ad Designifier (still BAD!).

I had a working version in about 10 hours and brought my laptop to dinner with our CFO and his wife because I was really proud to show it off. (”Wow, you really mean business tonight bringing that laptop.”)

I had some beta testers during Black Friday Cyber Monday (aka Thanksgiving) and then we rolled it out to the team the week after.

Big hit!

Internal tool for Brainjolt

Created the Brainjolt Ad Maker — a new AI-fueled creative tool in a manic 3 day sprint.

• H: “You did what my entire product team at ____ couldn’t do in 3 years.”
• Abraham: “You did in a weekend what ___ couldn’t do with 15 people over 3 months.”

That felt good to hear. It was adopted quickly by our traders and reduced time spent on creatives by 90%. Within a month we were spending hundreds of thousands of dollars a day on creatives made through it.

I rarely talk about Real Work Projects here, but work got crazy busy in Q4, and although I don’t like to spill All The Secrets, I’m very proud that we launched a super successful newsletter that quickly built up to a million subscribers!

In 2019 I had an idea to make a custom holiday video for each of our 25 team members. Then we thought … why do that when we could make the SAME video for everyone, but splice their name in a way that is obviously ridiculous.

To make it even stupider, we recorded (over Google Meet) Abraham and I saying it at the same time and trying to make our lips match in cadence. Thus was born (according to my journal) “the dumbest, most bestest holiday video ever made”:

Sure, it’s cringe, but if you don’t have a little fun then what’s the point of doing all this?

I’ve been building computer games for Z for a year now.

Thought I’d write up some stories and observations on it.

This was the fulfillment of my main goal of 2024: to find Brainjolt a new home.

I am so proud and happy for that to be Centerfield. They’re local to LA with a beautiful office in Playa. Their culture is so kind and positive, and they’ve created an environment that would simply be impossible for us to offer without them. It’s also a shock for us to go from complete insanity and planning by the day to creating an annual operating plan and departmental budgets.

This was the hardest project by far that I’ve ever worked on. It feels silly but there is no way I could emphasize enough how hard it felt. To balance running a company with such a lean team and high scale, while also trying to sell it… well, it sounds dramatic but my mind almost broke. It didn’t help that I was also running the foundation and investment fund.

The nice thing about this is everything else seems kind of easy now.

Now I’m adjusting to being in a more corporate environment and having bosses and accountabilities. I expect to learn a lot. I’m thankful for the opportunity.

AI-gen music has been pretty bad, but it’s JUST getting to the point of passable. Suno.ai is the best I’ve seen, so I decided to make a lofi youtube channel using it. In this video I walk through the process of the music generation.

We set up a 501c3 nonprofit because OH WE HAVE PLANS.

This is the output of a python mathlib script that connects to the Google Search Trends API (built using the help of ChatGPT of course)

• Inspired, of course, by After Dark’s Flying Toasters.

our beloved jigsaw puzzle company, born out of the pandemic

Over a decade of my life would be spent on viral media. At first we were 100% organic, with articles going viral a couple times a month from Facebook, Stumbleupon, Digg, Reddit. Then we got on the paid media treadmill and did a lot of content arbitrage. Slideshows and whatnot. Then after that was squeezed we moved onto the affiliate space, which culminated in an acquisition in 2024. In 2025 we surpassed all company records, selling a couple billion dollars of e-commerce products and becoming the #1 top publisher in the US for e-commerce affiliate sales.

iStockPhoto was an amazing place in the early 2000’s. It had a forum where photographers would critique each others work and chat about whatever. It was a supportive and creative community (mostly). I was always so proud when one of my photos was featured on the homepage.

I went from superfan to site administrator. At first it was more of a forum moderator role. A few people had god mode privileges — Bruce Livingstone (the CEO), some of the staff, and Peebert. Lol, peebert, if you ever read this you’re a crazy sonofa. I was a hardcore incredibly annoying evangelical Christian at the time and he was an atheist and we would go at it. He would be the one to threaten and be overjoyed to “banhammer” someone. He was a kind of cats paw for management … he did the dirty work, and I think we were all better off for it. Sometimes the power went to his head.

We all liked him, though, and I remember in 2003 when 6 of us admins all uploaded parts of a birthday message as photos and approved them in order so that a birthday message appeared in the “New Arrivals” section of the homepage:

Check out the stats section:

Total files 41,601 Waiting approval 584 Express lane 111 Applications waiting 8 Gigabytes today 9.84 Users online 28 Guests online 57

The “waiting approval” was a section where admins would approve or reject submitted photos. We had a whole list of rules — anything with logos/trademarks were instant rejects for instance — but some of it was subjective on if it was “good enough” for the site. We were paid $0.25 per image reviewed. A kind of mechanical turk before AI was available.

It was a formative part of my life. Bruce created a community of people who cared about something he built and made (a lot) of money from… not only cared, but would volunteer their time in support. (Something I would try to replicate in the future.) It fostered my creativity through photography … the kind that people would pay for. I would play and judge photoshop tennis, which improved my design and communication skills. It deepened my belief that you could have friendships and yet never meet in person; that you could run a company or community completely on the internet.

A final photo. I was such a fanboy that in July 2003 I took a screenshot of our 50,000th upload and still have it saved today. I don’t have much else from that period… but I do have this.

In 1998 I started playing around with a Kodak DC260 digital camera. Here’s what I think the first picture I took (or one of the first):

Gertrude the chicken!!! Such a sweetheart.

At some point I learned you could license your photos to strangers on the internet in exchange for money, specifically through iStockPhoto. Eventually the proceeds from that paid for me to upgrade to a Canon EOS in 2003. I was so proud of this that I took my first mirror selfie:

My (now extremely embarrassing) iStock profile from 2003 using Mozilla and you can see I’m chatting with Barb on MSN Messenger (which all the iStock admins used to chat):

If you were into Linux you had to have a solid Bill Gates jokes page (this is honestly so crude) … but also, this section (and the entire site really) is one of my first forays into being a “curator” … not a very good one yet, but it got me started.

In high school I was a TA for a tech class, which gave me just enough authority to try crazy things. It was never “part of the job” to create new learning modules … but, why not? I made one for Photoshop, which was straightforward because I already knew my way around it, and one for Blender, which I absolutely did not.

At the time Blender felt gloriously obscure: free, Linux-y, and the sort of thing that made me feel like I had stumbled into a secret society. I got it running on one of the lab computers, set it to boot straight into Blender. Success! Linux on a high school lab computer in a backwater Florida school where nobody even knew what Linux was.

I wrote up a tutorial on how to make a planet, printed it, then put it in a white 3 ring binder. It was focused on making a planet not because it was interesting, but because that was the one thing I could figure out how to make. Drop in a light source, add a sphere, wrap a texture around it, aim the camera somewhere vaguely cinematic and congratulations fellow high schooler, you made a planet!

Alas, I have no screenshots or photos from this. But I love that Blender survived and went on to become a real institution. I’m proud that, while still a kid myself, I got to introduce other kids to it.

As a teen I was contracted out as a local “webmaster”. One of my first commercial sites was a realty in my hometown of Venice, FL.

George got the clients. He’d charge them like $10,000 and pay me like $100.

You don’t want the person doing the webmastering work to actually get credit, though (it was kind of funny I put my name on there), so that was changed:

It all worked out, but I definitely wanted to be on the other side of the equation.

I’ll end with this incredible menu I made:

I put these posters around town:

This poster must have been a couple years after I first started it, because at first I used Wildcat! BBS on DOS. I tried a bunch of different BBS systems and learned a lot. Like ASCII and ANSI design, which is still so fun.

Later I migrated over to Excalibur BBS for Windows which was REALLY COOL at the time but introduced a lot of platform friction — the user had to login first in a terminal, then download the Excalibur Client, install, then login again. Still, it was new and fun and I really enjoyed designing in a GUI way.

Here’s a screenshot I found of Excalibur, it’s not mine, but gives you an idea of the look: