The art of the BEAM: How to replace the JVM and not get fired

The art of the BEAM: How to replace the JVM and not get fired

BEAM adoption through guerrilla style development.


  1. Show, don’t tell
  2. Ask for forgiveness, not permission

Andrei Soroker seems disappointed at the turnout to his talk - but lets face it; attendees of Code BEAM SF will tend to be staunch converts already - the people who need to watch this don’t even know yet that they need the BEAM in their professional life. But to the rest of us this talk is still entertaining.


I did guerrilla style development in my last job. I rewrote Python app to Elixir, deployed the Elixir one and turned off the Python one when no one was watching, and confessed to doing so few days later. Since then, at that job I was writing only Elixir :slight_smile:


I also did that. I wrote a new service in Elixir without telling anyone. The new service was not planned but needed to solve a problem and was a great fit for Elixir.

Well, nobody really cared and I am still not allowed to replace the other stuff with Elixir …


With that kind of approach to other people it’s inevitable that you’re going to have to face battles, if not wars. You won with the technology choice, but I feel something got lost… On the other hand, the Elixir got adopted at the company, whereas my 6 month long attempt that included 3 successful projects got killed off. Go figure.

I got Elixir introduced in a company I worked at. I tried a very logical process to introduce it, arguing how it was a better fit for a specific problem. In the end, I don’t think much of that mattered. When I finally got the okay, my manager said I trust you and the judgement you’ve had in the past. He didn’t say, you’re right, these three reasons are why we should use this. I think it’s hard for a manager, even a technical one to truly understand the differences. You really need to spend some time with each and get the feel of both to understand. In lieu of that, you need to build a relationship where people trust you. I think skunkworks projects are certainly the easiest way to introduce something, but I think it diminishes people’s trust in you. So, I don’t think it’s a good long term strategy.

1 Like