> Sure, but the harder part is measuring whether the developer is actually 5% faster. Otherwise you can make the same case for every $10/mo subscription service in the world, and so we should all be operating at infinite efficiency.

After n such iterations, the developer gets 100*(1-0.95^n)% faster. So, after some such $10/month purchases, the developer gets so fast that buying another improvement yields diminishing returns.