Thread by @amasad

We get this question a lot: “Do we build Replit in Replit?”

Short answer: yes, exponentially more so.

Longer answer has to do with the concept of bootstrapping. Thread:

https://twitter.com/nem035/status/1495850458996477959

When we started Replit we did not want to compete on features.

Feature races are boring.

Instead we wanted to be 10x better on some attributes (speed, collab, instantness).

That meant we innovate and we build something new. Something web native. Exciting, and fun.

That also meant we are okay with lacking some power features.

It also meant we are okay trading off scalability for speed & simplicity in the near term.

The result: we unlocked entirely new coding markets and use-cases.

Now, we *want* to add power as long as we can hold the entry experience constant.

Most systems trade ease for power, and hurt UX over time.

Just think of webdev in the early 2000s (notepad & html) vs now (npm, react, babel etc)

More powerful yes, but also hard to learn and use

Back to the question.

We use Replit more intensively at every stage of its evolution.

Our first use-case was a simple repl for sharing programs.

Then we used it for recruiting.

Then prototyping & collab.

Then bots and one-off services.

And now:

We are using it to build templates and environments in itself. Every language on Replit is fully bootstrapped.

We call this “repl space” (think userspace) and our goal is to move everything to repl space.

blog.replit.com/replspace-temp…

Internally we track our time spent in Replit and we like to see it go up month over month and year over year.

We follow Englebart’s bootstrapping ideas

Eventually the non-replit core will be so small we won’t have to change it for months.

Instead the system will be fully bootstrapped and everything is running repl space!