The birth and death of @searls

This is a copy of the Searls of Wisdom newsletter delivered to subscribers on March 1, 2025.

Congratulations on beating Level 2 of 2025. Can't wait to see what March has in store for us. A couple new initiatives to share from Searls Industries this month:

• Becky started a podcast in conjunction with her Better with Becky enterprise, where she sells monthly strength-training programs via a curiously-slick web application
• I announced POSSE Party, my first solo product. It will launch later this year, and its goal is to help people escape the post-Twitter diaspora by cross-posting your content to your social accounts on your behalf

Yesterday, I returned from Vegas after attending Test Double's annual retreat and quarterly board meeting. I'm coming to grips with a couple things. First, that I'm entirely okay never going back to Vegas. And second, that I'm entirely okay never going back to who I was the last time I needed a Vegas trip to escape my life.

Speaking of Vegas, I couldn't find a single picture from the trip, so let's all pretend this shot with my longtime colleague and collaborator Dave Mosher wasn't taken at Disney World a week prior:

On a personal note, I've changed in some pretty fundamental ways over the past 18 months, and—in case you've noticed how my character arc is distressingly similar to a bunch of other tech villains—I figured I'd explore my transformation in today's note. Yes, I moved to Florida. Yes, I started caring less how my words make people feel. Yes, I'm prioritizing my own wants and needs over those of others.

In short, I've learned to embrace my inner asshole. And I'm frustrated to report that it's totally backfired. Instead, I'm somehow a kinder, more generous, and all-around better person today than the whole last ten years I spent trying to be good.

The birth and death of @searls

Growing up, I had a keen intuitive sense that living for the approval of others would be stifling and counter-productive. More than most, I desperately yearned for acceptance and validation—the corrosive impact of which I was forced to reckon with at a very young age. In second or third grade, I remember playing H-O-R-S-E at my friend John's house and, as he passed me the ball, he asked why I cared so much what others thought about me. I didn't know the answer.

Content warning: more content…

Calling private methods without losing sleep at night

Today, I'm going to show you a simple way to commit crimes with a clean conscience.

First, two things I strive for when writing software:

1. Making my code do the right thing, even when it requires doing the wrong thing
2. Finding out that my shit is broken before all my sins are exposed in production

Today, I was working on a custom wrapper of Rails' built-in static file server, and deemed that it'd be wiser to rely on its internal logic for mapping URL paths (e.g. /index.html) to file paths (e.g. ~/app/public/index.html) than to reinvent that wheel myself.

The only problem? The method I need, ActionDispatch::FileHandler#find_file is private, meaning that I really "shouldn't" be calling it. But also, it's a free country, so whatever. I wrote this and it worked:

filepath, _ = @file_handler.send(:find_file,
  request.path_info, accept_encoding: request.accept_encoding)

If you don't know Ruby, send is a sneaky backdoor way of calling private methods. Encountering send is almost always a red flag that the code is violating the intent of whatever is being invoked. It also means the code carries the risk that it will quietly break someday. Because I'm calling a private API, no one on the Rails team will cry for me when this stops working.

So, anyway, I got my thing working and I felt supremely victorious… for 10 whole seconds. Then, the doubts crept in. "Hmm, I'm gonna really be fucked if 3 years from now Rails changes this method signature." After 30 seconds hemming and hawing over whether I should inline the functionality and preemptively take ownership of it—which would separately run the risk of missing out on any improvements or security fixes Rails makes down the road—I remembered the answer:

I can solve this by codifying my assumptions at boot-time.

A little thing I tend to do whenever I make a dangerous assumption is to find a way to pull forward the risk of that assumption being violated as early as possible. It's one reason I first made a name for myself in automated testing—if the tests fail, the code doesn't deploy, and nothing breaks. Of course, I could write a test to ensure this method still works, but I didn't want to give this method even more of my time. So instead, I codified this assumption in an initializer:

# config/initializers/invariant_assumptions.rb
Rails.application.config.after_initialize do
  next if Rails.env.production?

  # Used by lib/middleware/conditional_get_file_handler.rb
  unless ActionDispatch::FileHandler.instance_method(:find_file).parameters == [[:req, :path_info], [:keyreq, :accept_encoding]]
    raise "Our assumptions about a private method call we're making to ActionDispatch::FileHandler have been violated! Bailing."
  end
end

Now, if I update Rails and try to launch my dev server or run my tests, everything will fail immediately if my assumptions are violated. If a future version of Rails changes this method's signature, this blows up. And every time I engage in risky business in the future, I can just add a stanza to this initializer. My own bespoke early warning system.

Writing this note took 20 times longer than the fix itself, by the way. The things I do for you people.

Welp, I just did the thing I promised I'd never do and read the Hacker News comments for this otherwise lovely post pointing out the durable relevance of Ruby on Rails twenty years later. One comment stood out as so wrong, however, I couldn't help but clap back.

It started when someone wrote, presumably in praise of Rails, "I really like web apps that are just CRUD forms." CRUD is shorthand for the basic operations of "create, read, update, and delete", and those four verbs can express the vast majority of what anyone has ever wanted to do with a computer. It's why the spreadsheet was the killer app of the 80s and early 90s. It's why Rails 2's embrace of REST as a way to encapsulate CRUD over HTTP redefined how pretty much everyone has made web applications ever since.

Anyway, in response to the fellow above said he enjoys simple CRUD apps, somebody else wrote:

I really like easy problems too. Unfortunately, creating database records is hardly a business. With a pure CRUD system you're only one step away from Excel really. The business will be done somewhere else and won't be software driven at all but rather in people's heads and if you're lucky written in "SOP" type documents.

This struck me as pants-on-head levels of upside down. Father forgive me, but I hit reply with this:

As someone who co-founded one of the most successful Ruby on Rails consultancies in the world: building CRUD apps is a fantastic business.

There are two types of complexity: essential and incidental. Sometimes, a straightforward CRUD app won't work because the product's essential complexity demands it. But at least as often, apps (and architectures, and engineering orgs, and businesses) are really just CRUD apps with a bunch of incidental complexity cluttering up the joint and making everything confusing, painful, and expensive.

I've served dozens of clients over my career, and I can count on one hand the number of times I've found a company whose problem couldn't more or less be solved with "CRUD app plus zero-to-one interesting features." No technologist wants to think they're just building a series of straightforward CRUD apps, so they find ways to complicate it. No businessperson wants to believe their company isn't a unique snowflake, so they find ways to complicate it. No investor wants to pour their money into yet another CRUD app, so they invent a story to complicate it.

IME, >=90% of application developers working today are either building CRUD apps or would be better off if they realized they were building CRUD apps. To a certain extent, we're all just putting spreadsheets on the Internet. I think this—more than anything else—explains Rails' staying power. I remember giving this interview on Changelog and the host Adam asking about the threat Next.js posed to Rails, and—maybe I'd just seen this movie too many times since 2005—it didn't even register as a possible contender.

Any framework that doesn't absolutely nail a batteries-included CRUD feature-set as THE primary concern will inevitably see each app hobbled with so much baggage trying to roundaboutly back into CRUD that it'll fall over on itself.

Anyway, I believe in this point enough that I figured I should write it down someplace that isn't a deeply-nested comment at a place where I'd recommend you not read the comments, so that's how you and I wound up here just now.

I hate to spoil the end of this wonderful post for you, but:

In short, if your priority is developer experience, you won't find a better option than Heroku.

Heroku's greatest sin wasn't in selling out (in fact, its Salesforce acquisition was one of the least disruptive in the history of SaaS), it was in well and truly solving the problem it set out to address. Then, after that, …they didn't have much new and shiny to talk about. Everything Just Worked™.

There are a handful of rough edges that they're still sanding down—HTTP 2 support was the big one, but they've recently covered that. I take the fact that I reacted to their banner about a next generation paltform with worry it might make something worse as a GOOD sign. Those among us who actually give a shit about working software should do a better job of rewarding the handful of companies that pull it off.

(And if you do use Heroku, Judoscale's founder Adam is an old friend and a Searls-certified badass and you should check them out.)

Namanyay Goel with a post that appears to be lighting the Internet on fire and is making me feel pretty smart about emphasizing the generational transition at which the software world finds itself:

We're at this weird inflection point in software development. Every junior dev I talk to has Copilot or Claude or GPT running 24/7. They're shipping code faster than ever. But when I dig deeper into their understanding of what they're shipping? That's where things get concerning.

Sure, the code works, but ask why it works that way instead of another way? Crickets. Ask about edge cases? Blank stares.

Imagine if your doctor had skipped med school, but—don't worry!—the AI tool built into his Epic tablet gave him the right answers to every question that happened to come up during your intake consult and your first physical.

Hopefully you never need anything serious.

Turning your audio podcast into a video-first production

I was chatting with Adam Stacoviak over at Changelog a couple weeks back, and he mentioned that this year they've taken their podcast "video-first" via their YouTube channel.

I hadn't heard the phrase "video-first" before, but I could imagine he meant, "you record the show for video—which is more complex than recording audio alone—and then the audio is a downstream artifact from that video production." Of course, key to my personal brand is failing to demonstrate curiosity in the moment by simply asking Adam what he does, and instead going on an individual two-week-long spirit quest to invent all the wheels myself based on the possibly incorrect assumption of what he meant in the first place.

Anyway, as of v31 of Breaking Change, my podcast is now, apparently, a video-first production. I figured I'd share my notes on the initial changes to the workflow, along with links to the products I'm using.

Here's the video:

And here's the extremely simple and easy 10 step process that got me there (with affiliate links throughout):

1. Record audio and video in OBS
   • Video is recorded in 4k@60fps in 8-bit HEVC as an MKV file (because MKV files, unlike MOV, can be interrupted by a crash without losing the entire recording)
   • I use a Sony a7 mark IV over HDMI with Elgato Camlink 4K mounted via the Elgato Master Mount system and flanked on either side by Elgato Key Lights Air. I also have an Elgato Prompter in front of the lens that displays two windows, side-by-side: OBS on the left and my Things project with my show topics on the right
   • I record audio tracks from an SSL 2+ USB interface
     • Track 1 is reserved for the first XLR input, which has a bog-standard SM7b microphone plugged into it
     • Track 2 is L&R of the loopback interface for music and stingers (here's my guide on setting up SSL2+ for loopback that allowed me to avoid software like Audio Hijack)
   • Music and stingers are played manually from shortcuts on my Stream Deck XL
   • While recording, if I need a break, I only hit PAUSE/UNPAUSE instead of STOP/START to ensure only one file is created
   • When finished, leave it recording and then LEAVE THE ROOM for a minute to create some dead air I can later use to sample the room noise with iZotope's RX Spectral Denoise plugin

You'll never guess what happens next…

A script to validate videos for the Instagram API

If you are publishing videos via the Instagram API (as I do for my feed2gram gem and for Beckygram), one of the first things you notice is that it is a lot less forgiving than their app is.

From their docs that spell this out:

The following are the specifications for Reels:

• Container: MOV or MP4 (MPEG-4 Part 14), no edit lists, moov atom at the front of the file.
• Audio codec: AAC, 48khz sample rate maximum, 1 or 2 channels (mono or stereo).
• Video codec: HEVC or H264, progressive scan, closed GOP, 4:2:0 chroma subsampling.
• Frame rate: 23-60 FPS.
• Picture size:
• Maximum columns (horizontal pixels): 1920
• Required aspect ratio is between 0.01:1 and 10:1 but we recommend 9:16 to avoid cropping or blank space.
• Video bitrate: VBR, 25Mbps maximum
• Audio bitrate: 128kbps
• Duration: 15 mins maximum, 3 seconds minimum
• File size: 300MB maximum

If you get this wrong, you'll receive a mostly-unhelpful-but-better-than-nothing-error message that looks like this:

{
  "message": "The video file you selected is in a format that we don't support.",
  "type": "OAuthException",
  "code": 352,
  "error_subcode": 2207026,
  "is_transient": false,
  "error_user_title": "Unsupported format",
  "error_user_msg": "The video format is not supported. Please check spec for supported CodedException format",
  "fbtrace_id": "AvU9fEFKlA8Z7RLRlZ1j9w_"
}

I was sick of hobbling together the same half dozen ffprobe commands and then eyeballing the results (which are typically inscrutable if you don't know what you're looking for), so I wrote a script to test this for me.

For example, a recent clip failed to syndicate to Instagram and I wondered why that was, so I ran this little script, which I put on my PATH and named validate_video_for_instagram. It output:

Validating video: /Users/justin/Documents/podcast/clips/v30-the-startup-shell-game.mp4

✅ container
✅ audio_codec
✅ max_audio_sample_rate
✅ video_codecs
✅ color_space
✅ min_frame_rate
✅ max_frame_rate
❌ max_horizontal_pixels - Maximum columns (horizontal pixels): 1920 required; got: 2160
✅ min_aspect_ratio
✅ max_aspect_ratio
✅ max_video_bitrate_mbps
❌ max_audio_bitrate_kbps - Audio bitrate: 128kbps maximum required; got: 256.073
✅ min_duration_seconds
✅ max_duration_seconds
✅ max_size_megabytes

❌ Video had 2 error(s) preventing API upload to Instagram.
Docs: https://developers.facebook.com/docs/instagram-platform/instagram-graph-api/reference/ig-user/media

Is it surprising that the Instagram API won't accept 4K video? Yes. Especially since the videos weighs in at less than 100MB.

Want this for yourself?

To run this, you'll need a modern Ruby installed and ffprobe installed (on a Mac with homebrew, brew install ffprobe should do).

To be continued…