Coleton Gorecke (@coletoncodes) profile picture

Self Taught Software Engineer. I talk about life, software engineering, productivity, and growing your own solo software business. Building. Writing. Creating.

United States
Joined April 2018
Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)Photo by Coleton Gorecke (@coletoncodes)
I refunded a customer once and didn't ask why. Two months later I fixed the same bug three more people had filed. Refund requests are the cheapest user research you'll ever get. One question in the email: what made you want your money back? The answer is the next thing you build.
12
Everyone talks about engineers getting lazy with AI. Almost nobody talks about the part above them. Businesses have started pricing speed over quality, and the quiet truth is: quality has gotten less subjective. It used to be that "good enough" meant a person who understood the thing looked at it and said yes, this holds. Now "good enough" mostly means "shipped." And the incentive lands on you fast. The boss wants it yesterday. The model says the diff passes. So you merge, you move, you tell yourself you'll harden it later. Later becomes three weeks from now, at 2 a.m., when the shortcut you took brushes up against the shortcut someone else took. The next decade of bad software won't be written by lazy engineers. It'll be written by people under pressure to stop checking. You can't always stop the deadline. You can still be the one who checks.
16
AI didn't make us better engineers. It made us faster ones, and speed is the easiest thing in the world to mistake for competence. Pre-AI, you onboarded to a team, then dug into one part of the system until you actually knew how it ran. Four miles wide came later, but you had to earn the depth first. Now the model does the digging. You're four miles wide and an inch deep on day one. You can't feel it until it bites: a change that passed every test and broke a flow three weeks later. Speed was never the point. Knowing whether the thing is right was, and still is.
20
Replying to @everton_dev
It should technically behave such that on each load it fetches and caches the latest result, and then on the next launch it’ll read from the cache. So the cache is technically what is serving the config. This way you don’t suddenly jar your users with a config change while they are the middle of the app, but rather on the next “fresh” engagement of the app.
1
2
316
AI speeds up three things that used to be the whole job. 1. Reading the code you didn't write. 2. Asking why the thing is built the way it is. 3. Checking how two parts touch that call each other once a quarter. Each of those feels like the slow part. So the path of least resistance is to skip them and trust the diff. Then the once-a-quarter call happens, your change from three weeks ago is in the middle of it, and prod is down at 2 a.m. Speed offloaded the easy part. It made the careful part the only thing left, and that part's just you.
12
For years I rewrote old code for the sake of rewriting it. "This will be cleaner. This will be the way I wanted it the first time." Ego, mostly. One iOS app I rebuilt entirely because the business had shifted, and it genuinely was faster to rebuild than to onboard. Sometimes it's right. But a rewrite is a decision, and you can't make it unless you know where the bottlenecks actually are. You don't know that by pretending. You know it because you've been intimate with the product. AI makes rebuilding cheap. It doesn't make knowing whether to rebuild any less your job.
1
19
Coleton Gorecke retweeted
Another real App Store review. A planner should get out of your way. Today's Focus is a short list, on purpose. Join the live beta: intent-app.com/beta #productivity #minimalism #todolist
1
9
The thing I keep thinking about isn't that AI will take my job. It's that I'm quietly getting worse. Pre-AI, I had to know the framework or the feature didn't ship. If I wanted to build in a stack I didn't know, I read the docs, got the errors, sat in them until it worked. The only way forward was through the thing I didn't understand. Now I prompt, the model knows, and it works. I never had to learn it. So I ship more and understand less, and the scary part is how little I miss it. Speed has a cost nobody invoices you for: you stop becoming the engineer you were supposed to be. Using the tool well means not letting it be the reason you stop learning. The model can hold the syntax. It can't hold your growth.
17
Microsoft 365 was down for over a day, in 74 countries, two weeks ago. Not a hack. Not AI. Two network changes made months apart that were never tested together. One assumed how the other worked. The assumption was wrong. That's the thing AI can't save you from. It writes both changes fine. What it can't do is sit there sick at 2 a.m. wondering which path you just bolted onto another path and didn't pressure-test. Touching more code isn't the same as understanding it. The seam is still yours.
33
Someone told me software is a bicycle that upgrades to a motorcycle, then a smarter car. No. It's a city. One feature is a shed. Then another shed beside it, and a road between them. Then more buildings, more roads, until every new thing you add leans on roads you forgot you built. AI didn't make us better at this. It made us faster at it, and you build five roads in a day now instead of one. You stop noticing which roads are load-bearing. That's not a superpower. That's how you lose the plot.
19
Replying to @raroque
I write about this stuff as well, and speak about it with clients and colleagues. There’s a huge argument in today’s world on “people will just build their own software” as mentioned in Thariq’s post, and this is an excellent take. It’s probably a bit of self preservation as well, but I think it still holds true regardless. We didn’t stop going to the grocery store just because delivery exists (though we use that a lot) We didn’t stop paying for streaming services, despite a lot of this being possible with home media centers. There’s tons of other examples in probably spacing on but to your point on crowded niches/apps. My opinion has been, and always will be that ultimately a crowd means there’s demand, and you don’t really have to capture the entire market to be successful, but just a portion of it. Jira and Linear do the exact same thing as Monday or shortcut, but I prefer linear. I think it’s a mix of meeting your unique customer persona(s), and finding those users. Great perspective, Chris.
2
4
280
Coleton Gorecke retweeted
“good software is a set of strong opinions” My apps are in quite possibly the most crowded and played out markets (calorie tracking, todo, budgeting, personal crm) But that sentence alone is why people still pay for my apps and why there is always still space for a new competitor People aren’t paying me for a calorie tracking app, they are paying for my (sometimes strong) opinions on what a calorie tracking app should be
Replying to @trq212
this is one of those reasons why I don't think most people will really generate their own software good software is a set of strong opinions and I'd rather have something that's been really thought through by someone else
18
12
1
322
10,334
Coleton Gorecke retweeted
One task. 25 minutes. Nothing else on the screen. Done, not just busy. Join the live beta: intent-app.com/beta #focus #deepwork #productivity
1
9
You'll build things that suck. You'll make decisions that suck. You'll fail your own expectations over and over. And every so often, you'll get a moment where you think "holy shit, this worked." That moment is never the end. It's the beginning of the next hard thing. Entrepreneurship is a long series of failures with brief wins in the middle. The rollercoaster isn't for everyone. You have to be strong enough to keep pushing while the thing grows. And it grows you as you grow it.
27
Passion for the idea is optional. Passion for the standard isn't. There will be a thousand things on this path to solopreneurship you aren't passionate about. Taxes. Marketing. Customer support at 11pm. Do I like those? Fuck no. But they're the price of getting good, and getting good is the thing that actually pays. A side project runs on love for the idea. A business runs on love for how well you do the parts you hate, and how well you serve your customers.
27
Replying to @fortranibal
I’ve had multiple jobs basically since I was 18, but got really serious about turning side projects into real income in the last 5 years.
1
3
Coleton Gorecke retweeted
A parent's Tuesday: fourteen things, all shouting. Three of them matter today. Everything else has a place. Join the live beta: intent-app.com/beta #parenting #momlife #dadlife #productivity
1
10
You don't need talent. You need to show up and not quit. From 2015 to now, I 10x'd my income. No school for this, no real talent. I graduated from a high school with 50 kids in the class. All it took was finding high-paying skills, getting as good as I can at them, and refusing to stop improving. The only limiting belief is thinking you don't deserve great things. The price of admission has never been ability. It's consistency.
24
Most people picture building a business or holding a job until they can "retire" someday. I think that's backwards. A solo business isn't a retirement plan. It's the rest of your life. The problem in modern America was never work. It's that we got used to working for things we don't own, so our jobs end up shaping who we become. If you can build, go build something you own. The alternative is letting someone else own your time.
20