Jarno Virtanen

Fair point!

Why not listening to Steve Reich?

Family includes everyone! ❤️

In the end, your approach to handling increased load and any type of outside pressure is effectively determined by this power dynamic between the upper management and the chain of command. Thus, I would suggest, it would be a corollary to the Conway's law.

Whatever seems to work, at least for a while, say over-provisioning of resources, gets prioritized instead of the normal exploration of bottlenecks and understanding the system in a holistic manner. You're not rewarded for understanding, but for simplistic solutions that work in the short term.

When things are going smoothly, we at the lowest layers of hierarchy get to decide how we operate the systems. When failures pile up, the upper management starts to increase organizational and inter-personal pressure downwards and this results in efforts to please the displeased upper management.

A corollary to the Conway's law should state that in addition to shaping the architecture of (software) systems by structuring the company, the upper management also determines the approach to handling increasing technical demands by applying direct pressure to the mgmt hierarchy below.

And it's not like going to gym or a run in that it might feel awful during and great afterwards. Sauna feels great during the sauna session too. At some point it might start to make you feel bad and then you either stretch it bit based on your own preference or you go out of sauna.

Reminder to all non-Finnish people: going to sauna is a reward, not a punishment. It might feel awkward and even distressing at first, sure, but eventually you go to sauna, because it's one of the best things you can do in life. If going to sauna didn't make me feel great, I wouldn't go there.

always funny seeing this one come up in the wild... i made this in the bathroom of my job in 2018. however it took me another 10 months to quit that job because it is hard to take your own advice

I see nothing in the this new world of more AI-generated code that would reduce surprises or accidents. Everything points to the other direction. The lessons from resilience engineering I think will become more relevant over time.

I'm not worried that changing the implementation or adding new features will become harder. I'm worried that when accidents happen, we no longer have the graceful extensibility and the ability to circumvent the problem to resolve incidents involving several components or sub-systems.

We did a thorough post-mortem investigation and concluded that the root cause of the incident was: capitalism.

Action items: destroy capitalism.

The concept of desynchronization is such a great way to frame the problems of automated code generation. It goes a long way toward explaining why we see huge productivity increases for individuals working on greenfield projects, and modest (if any) in bigger projects over longer time frames. 1/6

Oliko edes kääritty lahjapaperiin?

Ok, I was probably too quick to praise the new version. While there are improvements like better speed control and the overall speech quality, it appears the speed drops to normal (from say 1,75x) from time to time for no apparent reason.

(Listening to web pages also reveals how awful many of them are for screen readers.)

I use the iPhone's built-in screen reader to listen to web pages and blog posts. You can start the screen reader by swiping with two fingers from the top of the screen downward. (It takes a while to hit that particular gesture.) For my surprise, it has gotten better in the iOS 26!

A completely new set of people will find out that making computer programs of few hundred or thousand lines of code is very easy and making computer programs of hundreds of thousands lines of code is extremely difficult.

AI seems well-suited for cyberattacks, because it doesn't matter if the AI hallucinates or gives the wrong "answer" so long as any _one_ of those attempts is correct. And furthermore, you will know right away if the AI has given the "right answer", a successful breach or any other foothold.

Read the paper! It doesn't say what the internet thinks it's saying. And what it says is very flimsy and could mostly explained with for example regression to mean.

It annoys me to no end that people who refuse to look anything other than meta-reviews when it comes to scientific evidence are perfectly willing to spread around the caricature version of Dunning-Kruger without even cursory look at the original paper, let alone demanding that it should replicate.

Welcome!

Okei, tässä on vähän liikaa Howeyn henkilökohtaista elämänfilosofiaa. On sillä jotain mielenkiintoisiakin pointteja, mutta jooh.

The Knowledge Project -podcastissä oli muuten aikoinaan mielenkiintoinen keskustelu itsejulkaisemisesta Hugh Howeyn kanssa.

Hyvää tarinointia ja aika vähän vaahtoamista (mikä on hyvä).

Kirjallisuuden puolella ei taida olla "bandcampejä" mistä voisi ostaa kirjan niin että rahat menisivät artistille suurimmalta osin ellei sitten kirjailija itse mene täysin itsekustannus/julkaisulinjalla?

I need something like STAMP (Systems-Theoretic Accident Model and Processes) for my personal life.

I get the feeling from the AWS public post-mortems that they do fixate on finding one root cause and this (at least partly) contributes to the fixes being more introducing automation and there being the disregard to the concerns you raise in the blog post.

I'm waiting for companies to roll out premium subscriptions that allow you to turn off AI features.

My favorite place in the world.

I hope the water is getting colder also. That said, I guess it's pretty chilly already, around 10-12 degrees?