Diego

software dev

[Diego @DiegoCode_]

@nicknisi `extraordinary circumstances` is not something you want to hear while on a flight lol

[Diego @DiegoCode_]

@AdamRackis Sheesh

[Diego @DiegoCode_]

@cmgriffing @modretro @OpenAI That thing looks like fun!

[Diego @DiegoCode_]

@orcdev My agent is

[Diego @DiegoCode_]

😢🫡

[Cristiano Ronaldo @Cristiano]

Portugal estará sempre em mim

[Diego @DiegoCode_]

I just completed "Inverse Captcha" - Day 1 - Advent of Code 2017 https://t.co/rjACGXGpGW

doing #AdventOfCode with @andrewhoyer group doing warm up for 2026s adventofcode. lets see how far i make it this year

[@levelsio @levelsio]

Guys literally only want one thing and it's ground beef and rice

[Addy Osmani @addyosmani]

https://t.co/4Hfk0dYIBI

[Aaron Francis @aarondfrancis]

The more pain you can eat on behalf of your customers, the bigger your moat

[Dillon Mulroy @dillon_mulroy]

one of the best talks i’ve watched in past few years

[4 of mallace @IceSolst]

“Patch the planet, we can fix everything” is one of the most misleading, and harmful security initiatives I’ve seen.

As if spending billions of dollars cooking the oceans to run swarms of agents spinning their wheels applying hourly? minutely? patches on every abandoned instance across the planet, racing against every incompetent sysadmin - no you will never catch up with human stupidity.

And it’s completely the wrong direction. It’s this old school mentality that “I’ll stay in my IT lane, and just update every package and infra” as if this will ever amount to anything. As if this hasn’t been the strategy for half a century.

Obviously you’re supposed to patch. But why be a hamster in a wheel spinning idly, while incompetent devs slop-cannon millions of lines of some of humanity’s most atrocious code ever conceived, bypassing every security recommendation, every known best-practice, to meet an arbitrary deadline set by their private equity overlords to ship some inane b2b saas that nobody is going to use, but will get everyone’s data leaked.

You won’t patch everything. You won’t catch every bug. You know it’s stupid, it’s a helpless strategy because you don’t want to address the real root cause: making software that doesn’t suck ass. Engineering principles that aren’t guided by psychotic greedy men in suits that just want a return on their investment at the cost of everyone’s data.

The problem is nobody gives a shit about breaches. “Cyber insurance covers this”, you send the breach notification, everyone shrugs as if it’s the accepted cost of living in society.

So companies have no incentive to care. Some of the most incompetent passionless people become “engineering leaders” that “understand the business” of shipping atrocities, without an ounce of consideration or thought, pure monkey level poop-slinging.
The worst code you’ve ever seen is production code, self-merged by the guy thinking increasing shareholder value is how you barely make it across the retirement line.

“Move fast and break things” has been one of the most dangerous principles, and an embarrassing one to hold today.

We have solutions. The dumbest dev can still ask a coding agent to find, explain, and address security issues in software architecture, and in implementation. And it does a better job than the majority of security engineers I’ve worked with.

But they do the opposite. They go AROUND and bypass every security recommendation anyone or anything makes. I’ve seen Claude-generated comments in code saying “this is highly insecure but we have decided to accept the risk” for things with trivial fixes.

God forbid, you’re going to have to work with devs. Teach them software design, architecture, planning how components and microservices interact, how to make safe code, how to keep it safe, ultimately just to be considerate.

To care a little bit about your craft.

“Patch the planet” is a waste of time. Invest in what we ship instead. It’s the ONLY way to make software resilient.

[Diego @DiegoCode_]

@leonnoel I also got an email for x money relevant ads. But that 6% apy is tempting.

[Diego @DiegoCode_]

@liran_tal @bot remind me of this for later

[Diego @DiegoCode_]

@thdxr Very nice! Gonna try this as I’m taking screen shots but then going to the chat to send it didn’t know you could send it directly 😎

[Diego @DiegoCode_]

@theo @bot why is this a wild read keep it short

[KJ Hardrict @KjHardrict]

right now people either trust AI way too much to NEVER make mistakes, or don't trust it at all after it makes ONE mistake.

in September CNN put out a story about an AI chatbot providing our military with hallucinated info during the Iran war.

an intelligence analyst had been using AI to help write up a report on a ship, but somewhere in that process the chatbot got the ship's cargo wrong, and the report it produced falsely claimed that a Chinese vessel in the Middle East was carrying parts for a nuclear weapons program.

but the analyst trusted what the AI gave them and submitted the report normally, and it moved through the military chain the way a report of this magnitude normally would.

nobody questioned where it came from until right before they were planning to board the ship. that's when somebody took a harder look at the report, realized an AI had written it, and shut the operation down.

to me, that's over-trusting AI at its absolute worst.

but a story like this usually pushes people too far the other way, to where they stop trusting AI at all.

if you look at what actually went wrong, though, the chatbot made one mistake, and the bigger failure was that nobody checked its work until it was almost too late.

that's why i think people need to learn how to afford AI the grace they afford themselves. when you get something wrong at work, you don't decide you're useless, and you never stop checking your own work.

the model is going to be right on most things and wrong on some, the same as a person. the difference is that it produces so much, so fast, that its mistakes add up quicker when nobody's reading.

getting the best results out of AI is just as much a human process problem as it is a technical one.

[Diego @DiegoCode_]

Life maxing

[Diego @DiegoCode_]

@DThompsonDev Love it makes total sense. Can’t wait to watch the pod!

[Diego @DiegoCode_]

@thdxr Why does OpenCode mobile have a location feature ? lol