Hacker News Reader

Comment by goodpoint | Hacker News Reader

> I don't want to spend my leisure time fighting with tools, debugging build processes that break all the time, make sense of docker's tagging system

Now that CV-driven development is the standard, we are forced to learn new bloated "devops" tools and services every year.

This type of knowledge expires very quickly.

Also these tools and services do not develop communities: business users drop last-year tools like hot potatoes when a new cheaper or shit^ny one comes along.

This hype-fueled corporate-driven culture is not going to end well.

View on HN

Sometimes I wonder if modern corporate culture has adopted soviet style newspeak.

Why are we immigrating to the new thing? For the same reasons as we migrated to the old thing. Why is the new thing better? Because it’s just so much better.

Because there are people for whom "playing" means implementing new devops tools, cluster managers, application servers, etc. In production.

If you can convince the higher ups that migrating tooling provides business value, you can do something that requires zero insight or creativity for a few months and at the end of it be promoted for organizing a successful migration despite delivering zero measurable business value. The effects of the migration are gonna be second and higher order effects that are impossible to separate from the rest of the business so you can just claim the tooling was the leverage that let the people doing first order work succeed. Being a tool astronaut lets you claim a portion of everyone else in your orgs success without ever taking a professional risk, since your odds of being called out for an ineffectual change are near zero, since there’s no first order signals.

I’m not saying modern tooling is useless; I don’t use ed, cc, and make for my development. But there’s a huge difference between a zero to one tooling effort and an N to N+1 tooling effort. The first one requires figuring out all the implicit/implied/manual parts of the process. The second one is often just turning one set of configuration languages into another.

Holy cow. This! Exactly this! It happened in big corp., I am working at right now. As an engineer, I just watched it happen. When I tried to explain to other engineers, most didn't get it. Only a few got the trajectory we were put in by higher ups for credits they were aiming to get. But those few were onboard or powerless as me.

We just observed it happen. As the engineers, we did our job best we could regardless. Things broke, we fixed. Somethings rolled back, and even for those they claimed credit & celebrated for re-inventing the wheel.

> huge difference between a zero to one tooling effort and an N to N+1 tooling effort

Spot on. There's a serious point of diminishing returns when piling up tools.

At some point the cost of learning, adapting, deploying, maintaining yet another tool becomes a net negative.

Often the solution is to replace 5 fancy tools with 100 lines of Python and Bash.

People on HN are always deriding the “tech hedonic treadmill”, but I find it interesting that this meme seems to have reached maturity in 2015, and the list of things they point to as “the new trendy thing this year” has been the same ever since: Some combo of Node.js, React, webpack, Docker and Kubernetes.

I think this meme was still fairly true then: Node was at risk of splitting into Node.js and IO.js, React was still new-ish and people were (rightfully) panicking over the drastic changes coming in Angular v2, Docker was still new-ish and there were arguments over whether Docker Swarm or Kubernetes was the future. But nowadays Node has stabilized, React and webpack have been comfortably dominant for years, Docker’s model of containers won out a long time ago and Kubernetes is there if you need it, but is mostly useful for people building a PaaS.

If people don’t want to learn these things, then fine, but I do wish the overall mood here would update to reflect the fact that we are now living in a much more stable time than when these memes were created.

> we are now living in a much more stable time than when these memes were created

Quite the opposite. The cambrian explosion of tools and SaaS is getting faster and the average useful lifetime of anything released today is getting shorter.

New hypothesis: People like to move on to the next thing once they have mastered the current thing, so in a weird co-evolution, those tools-of-the-year evolved to not be masterable and thus people haven't been able to move on yet.

Also, you shall shortly be smited for not having included Rust in that list. :)

Reply on news.ycombinator.com