Ruby/Rails: The Once and Future Platform (But Not Now)

As I chose a stack for Hi Marley Labs, Rails was not a part of my decision framework. That deserves discussion.

I love Rails. It shaped my journey as a software craftsperson.

I found Rails in 2006. The community was amazing. I devoured all 400+ episodes or Ryan Bates’ Railcasts. And I believe David Heinemeier Hansson is one of the most brilliant open-source architects.

Rails was magic. I built my first company, Pipewave, with Paul Boruta on Rails/Heroku. Thanks to Rails, I did this with a micro team versus my competitors who were fifty-times our size.

The Rails community taught me that “Code is Craft,” thank you: Dave Thomas ‘s Pragmatic Programmer (still essential reading).

Everyone who is serious about their craft as a software developer owes it to themselves to build and deploy a project with Rails.

Over the last decade, I moved away from Rails for a handful of reasons (the advent/fashion of SPAs, acquiring companies being either python or node shops)

There are many things I love about Rails, but one of the main things is that it’s ON THE RAILS—it makes a lot of the decisions for you, so you do not have to. And they are great decisions:

• ActiveRecord: the ORM all others aspire to
• ActiveJob: just the right level of complexity
• ActiveStorage
• ActionMailer
• A Great MVC paradigm
• Oh and JetBrains RubyMine is an amazing IDE

So, when I started to think about my stack for Hi Marley Labs, I really wanted the answer to be Rails.

But, I am going with a Typescript stack.

Here is why:

I expect the team working on Labs to be 30x engineers, and the 30x will come from leveraging LLMs to help with our coding. And this might just be a moment-in-time as LLM coding agents are much better at writing JavaScript than Ruby. And that is a big deal.

JavaScript is the dominant language on Github, followed by Python. Ruby is pretty distant. And therefore, generic coding LLMs will just be better with JavaScript.

That, plus, with LLMs doing much of the work, a typed language becomes critical (efficient tests, lining, and typing are critical in this new way of working).

I really hope that a year from now Ruby/Rails will be the answer again. And I do think that the Ruby community is amazing.

This is what I would love to see:

• A custom model focused on Ruby/Rails. How amazing would it be for each version of Rails to have an LLM that we can run locally to help with our coding? Rails has a huge advantage, being such an opinionated framework.
• We need a new Heroku. Vercel and Supabase have eclipsed Heroku/Rails as the battery-included platform.

For more insights and to share your thoughts, follow me on LinkedIn!