SuperGeekery: A blog probably of interest only to nerds by John F Morton.

In Part 1, I set up Google’s Gem­ma 4 run­ning local­ly on macOS with Olla­ma and cre­at­ed a sim­ple gem­ma shell func­tion. It worked, but it was bare-bones — no auto-start, no mark­down ren­der­ing, and the inter­ac­tive mode was just a pass-through to olla­ma run. Here’s how I turned that basic short­cut into some­thing that feels like a pol­ished chat inter­face, all with­in a sin­gle shell func­tion. The Prob­lem with Plain Text Gem­ma’s respons­es… + read on

Large lan­guage mod­els don’t have to live in the cloud. With Olla­ma and a recent Mac, you can run Google’s Gem­ma 4 mod­els entire­ly on your own machine. No API keys. No usage lim­its. No data leav­ing your lap­top. Here’s how I set it up. Installing Olla­ma Olla­ma is the eas­i­est way to run open-weight LLMs local­ly. On macOS, it’s a sin­gle Home­brew com­mand: brew install olla­ma Once installed, start the Olla­ma ser­vice: brew… + read on

I’ll be hon­est: when AI com­pa­nies start­ed scrap­ing the web to train their mod­els, my first reac­tion was frus­tra­tion. Years of care­ful­ly writ­ten con­tent, hoovered up with­out so much as a heads-up. But over the past year, my POV has shift­ed. Con­ver­sa­tions with devel­op­er friends, arti­cles pop­ping up in my feed dai­ly, the emer­gence of the llms.txt spec­i­fi­ca­tion — it all point­ed in one direc­tion. This isn’t a pass­ing trend. Peo­ple are increas­ing­ly skip­ping… + read on

When I shipped Bespo­ken 5.2.0 ear­li­er this month, the focus was on refin­ing pro­nun­ci­a­tion rules, mak­ing it eas­i­er to strip unwant­ed char­ac­ters from nar­ra­tion scripts and giv­ing you visu­al feed­back in the set­tings UI. That release laid impor­tant ground­work, but I knew the plu­g­in still had a hard ceil­ing: if your entry text was too long, Eleven­Labs would reject it out­right. And if you were run­ning a mul­ti­site install, things could break in sub­tle,… + read on

When I first start­ed in adver­tis­ing, I worked with an art buy­er and pro­duc­er named Karen Meenaghan at Ammi­rati & Puris. (I loved hang­ing out in her office.) She intro­duced me to an idea that has stayed with me ever since: Good. Fast. Cheap. Pick any two. It was framed as a fun­da­men­tal con­straint of the uni­verse. It wasn’t advice. It was more like the laws of physics. If some­thing was good and fast,… + read on

I recent­ly pub­lished a new plu­g­in for Craft CMS called Login Lock­down. It mon­i­tors login attempts on your Craft CMS site and, if failed login attempts sur­pass a thresh­old you define, the login process from that IP address is locked down for a spec­i­fied amount of time. You can con­fig­ure the plu­g­in to noti­fy you about lock­outs (via email and/​or Pushover noti­fi­ca­tions) or let the sys­tem run silent­ly on its own. The con­trol pan­el… + read on

This post will show you how the search page on this site works and how you can build the same thing your­self in Craft CMS. What makes the search page inter­est­ing enough for a long blog post? The short answer is that it func­tions like the search you’d expect to be built with a big fron­tend frame­work like React, Vue, or Svelte. Instead, it’s built with a tiny 10.7kb javascript library called Datas­tar. The… + read on

Today, I final­ly made a voice clone of my own voice to use on this blog. I had actu­al­ly done this process once before using the pro­fes­sion­al voice cloning option from Eleven­Labs. But I didn’t end up using that ver­sion — it just didn’t sound quite right to me. The mis­take I made was my own fault. To cre­ate a pro­fes­sion­al voice clone you need at least 30 min­utes of clean audio. I dis­cov­ered that with­out… + read on

Text-a-Mess is a basic web app that does one thing: It mess­es up the text you enter by adding dia­crit­i­cal marks through­out to obscure it while keep­ing it some­what read­able visu­al­ly. It does not encrypt the text in any way. A sim­ple text search will still find the orig­i­nal text. How­ev­er, it will severe­ly mess up any text-to-speech soft­ware because it will use dia­crit­i­cal marks to alter text pro­nun­ci­a­tions. Please use it with that in mind. Why… + read on

The col­or scheme of SuperGeek­ery depends on the day you read this blog post. If you don’t like the col­ors you see today, return tomor­row, and they’ll be dif­fer­ent. This post will show you how I gen­er­ate each day’s col­or scheme. It’s based on a sin­gle col­or cho­sen based on the day of the year. I then use CSS’s rel­a­tive­ly new OKLCh col­or func­tion to gen­er­ate 20 pos­si­ble col­ors through­out the site. The Oklchro­ma site inspired me to… + read on