Take every piece of software and OS, and you will find conflicting views of whether it goes faster or slower with time. You buy a new device, everything feels s

Take every piece of software and OS, and you will find conflicting views of whether it goes faster or slower with time. You buy a new device, everything feels snappy, but one year down the line, things feel slow. But the problem is, this is based on expectations, based on hardware, based on usage pattern, etc. What we need is a objective software speed measure based in UI terms (because that is what people in the end notice). Such as "time to start" and "time to get to usable state". This should be measured automatically after each build, and normalized to hardware. Only then can both developers and users see hard evidence that software, e.g. Chrome, is now running slower than before. Because the cost of adding a feature is otherwise invisible to the developer, while the user's keep getting into anecdotal discussions of what is faster or slower, comparing wildly different hardware capabilities.