Hacker News Reader

Comment by s-shellfish | Hacker News Reader

Pressure to compete. I don't think it's anything companies convinced people into doing, it seems more like one of those unwritten rules. Especially in tech. Your system breaks, you are responsible. Not good to simply look at the clock and say 'well that's a day' if you want more responsibility, leadership, respect, trust etc.

View on HN

> Not good to simply look at the clock and say 'well that's a day' if you want more responsibility, leadership, respect, trust etc.

Sadly, the effect of this is that there's less incentive to build resilient systems, since there's always someone available to fix things quickly. Trying to convince corporate to spend more time and money on building a system that looks/smells/feels the same, but is presumably more resilient, is very very difficult.

> Sadly, the effect of this is that there's less incentive to build resilient systems, since there's always someone available to fix things quickly.

Yep. It's often double-bad end result, since if you do build a resilient system, no one sees you "fix" it, so it looks like you did no extra work.

But if you build something broken, and then swoop in and "fix" it, you get extra credit for pulling "more" work after hours / during an emergency.

> Sadly, the effect of this is that there's less incentive to build resilient systems, since there's always someone available to fix things quickly.

I found this to be the opposite as long as the developer who wrote the system also owns the running of said system. It will not take many late night/weekend calls before the developer makes it more resilient.

Selling this to the business can be easy, because even if someone is there to fix it the system was presumably still down. This is why I push every developer to learn communication skills, and rudimentary business skills. If the developer (or their manager) cannot communicate to the business why a more resilient system is better for the business that's on them. This also requires developers to come off the "I only want to do it perfectly to 999999s", and again think about what can the business reasonably afford to do.

What they do is make it resilient to multi-hour fixes, while still keeping the light issues that look like emergencies but can be fixed by running a bat file to still go through, minimizing the work they have to do personally while still making them look like a hero.

Of course this will fall apart once their manager catches on but in some monolithic organizations that could take years

Sad but true. As I climb the ladder, all the decisions management use to make that I use to regard is idiotic make a lot more sense to me.

It's so hard to predict which projects will succeed and fail. People will spend time setting up Jenkins pipelines and all kinds of unit tests, and their stuff still breaks. I argued for unit-testing on our current project to add resiliency, and I swear 70% of my energy is fucking unit tests now. Whoops!

Resilient systems will still break. They may break less, but it's impossible to anticipate every possible usage permutation of a system from here until eternity. There are tons of reliable methods to build resilient systems and they can still have their faults. What I'm talking about is resilient system or not - someone has to be there to fix it if it breaks. There is no silver bullet when it comes to software - sometimes the difference really only can be who puts the most time in.

Minds require rest to function optimally, so pushing yourself to an extreme comes at a cost that gets realized down the line instead. Identifying which means of functioning comes at the greatest cost is not easy to analyze or ascertain. So working together in addition to competing is important as well.

All principles of capitalism. Just, never before in history has there really been so much attention paid to the mind. Software, eesh, yep.

This is precisely why it’s difficult to sell upward – why spend time and money on something, yet still not be able to guarantee that things will work? Especially when you can roll the dice, have things work great until such time it doesn’t, and then just throw some man power on it till it’s fixed. When thinking about cost, it can be really hard to justify something you can’t really see the benefits of, until after the fact.

Reply on news.ycombinator.com