Brooklyn

I’ve been thinking about how, in software engineering especially, it feels like the balance of who is the builder and who is the user is shifting. It used to be the case that the software engineer was clearly the builder. I don’t find that to be the case anymore.

Being the builder is fun. You get to try new things, experiment, and see your work come to life. You have a lot of control over the process of how you work.

Being the user, especially of a rapidly changing product, is not always fun. As a user, you adopted a product initially because of its utility, but then the changes start.

Pricing changes. Features evolve or go away. The builders try and increase the scope of their user base.

The tools used for software engineering are changing. Language models are in everything, for better or worse.

As software engineers, we’re now experiencing what many of us have put our users through for years. But it’s not being done by any one company, but by our industry itself to us.

More than I can ever recall, organizations seem to be measuring statistics like

  • lines of code written
  • PRs merged
  • issues closed
  • daily active sessions on internal tools
  • (and now) inference tokens used

Engineers are now the targets of the optimization loops, previously mostly reserved for the users of the products engineers built.

Prodded with surveys

On a scale of 1 to 10, how likely would you be to recommend this product to a friend or colleague?

Turned upside down and shaken for a bit of knowledge to feed into the tool makers’ road maps.

I get it. Language models change the game for builder software.

I feel this way.

The tools are useful. The demand for them is real.

But this post from Lenny has stuck with me:

Frantically.

Organizations are pulling out all the usual tools to try and make things happen. And engineers are feeling the pressure.

raw edit commit