Comment by twic | Hacker News Reader

Comment by twic | Hacker News Reader

Knowledge is imperfect and uneven. People don’t automatically know what should be normal

This has been a big one in my experience. Because our industry is growing fast, at any point in time, most people have not had very much experience. I worked a company which had hired a load of bright young things straight out of university and trained them up. They were great people. They also had no idea that a lot of what they were doing was really bizarre and unproductive. Or, if you hire people who have had a few jobs, they're often not very varied: for all values of 'Rails', someone with five years of moving around doing Rails will know a lot about Rails, but won't be in a position to appreciate just how screwed up a lot of things in the Rails world are.

It's hard to escape from this trap, because even when you look outside your office, there's so much duff information available. You read an enthusiastic article about CORBA/XML/model-driven architecture/SOA/MongoDB/microservices/lambdas/blockchain/etc; as someone with a few years' experience, in which time you've never touched any of these things, how on earth are you supposed to be able to work out that it is in fact terrible advice?

Less ageism? Less pop culture? Fundamentally, less fun and more seriousness?

Because it's so much fun to create new things in our industry, so much new stuff gets created. Often it's only marginally better than what came before, or more frequently, it's only better in one dimension, and significantly worse in other dimensions. But it gets adopted because it has social currency, recency, buzz, etc. and all the deficiencies slowly get filled in. And by the time it's mature, something else new comes along.

As long as we keep hiring cheap kids to build our stuff, they'll keep reinventing the wheel, and the workforce available to hire will only understand, and more importantly only want to work with, the latest popular fads.

A big reason for the bias towards new things is that greenfield development is easier and more fun, especially in organisations that aren't capable of producing good software. Learning a new framework every couple of years makes sense if it means you get to avoid having to work on confusing spaghetti code from a decade ago.

Ive been depressed professionally lately because I realized the practices that get implemented about half the time today are things my cohort and I were pushing for 10+ years ago as being essential to sanity. With such a long window for adoption it means you get to have the same six arguments again and again and again.

I really can’t atomach the idea that my job is to solve the same set of problems over and over again.

I may have to convince someone to hire me in a specialty field I’ve never worked in before just to fight the tedium and to trick myself into thinking these are new problems.

By learning principles. Always ask yourself, "How does the rubber hit the road?" And remember that every layer you add adds performance overhead and further opportunities to have trouble debugging.

Software development is all tradeoffs. Understand the tradeoffs, and you'll understand the tradeoffs inherent in a new technology. There really aren't that many new things under the sun.

Of course this is easy to say, but most don't do it. And virtually every new shiny is chock full of bad ideas in addition to the good ones. So after a while you'll become naturally cynical...

Reply on news.ycombinator.com