Giving More Tools to Engineers — Latent Demand and the Productivity Inequality

Erik Bernhardsson argues that software engineer productivity improvements add up on a logarithmic scale — orders of magnitude over decades. And unlike diapers, demand for software is not fixed.

The latent demand model

When the cost of building software goes down, three things happen simultaneously:

1. More software gets built
2. Engineer salaries go up
3. The number of engineers grows

This is counterintuitive. With fixed demand (like diapers), productivity gains cause layoffs. But software has massive latent demand — things that weren’t worth 1,000 hours of engineering effort become worth doing at 100 hours.

The productivity inequality

This creates a positive feedback loop where lagging companies fall further behind:

• Lack of tool adoption → falling behind peers
• Higher market salaries → priced out of top talent
• Failure to reorganize the factory → lower iteration speed
• Lack of engineers → temptation to adopt no-code tools with hidden costs

A “different kind of hard”

Engineering today is not easier — it’s a different kind of hard. Foundational problems are “mostly” solved, but knowing which off-the-shelf tools to stitch together based on best practices is its own deep expertise.

Connects to data team operations, leverage, tools and infrastructure.

Open questions

• Does this same latent demand model apply to data/analytics teams? As analytics tooling improves, does demand for analysis grow proportionally?
• Where are we on the logarithmic productivity curve for data teams specifically?