[kkowalczyk]

So if we apply your logic to software, a "thoughtful R&D process" of yearly releases (also known as "waterfall software development") is better than agile process of shipping frequently and tweaking your future features based on customer feedback.

Etsy (http://codeascraft.com/2011/02/04/how-does-etsy-manage-devel...) and just about any web dev shop disagrees with you.

There's nothing different about hardware.

In fact there's a story about a guy who build human-powered airplane not by planning harder but by increasing the frequency of iterations (http://www.fastcodesign.com/1663488/wanna-solve-impossible-p...).

This is exactly the idea behind Xiaomi's hardware and software efforts: improve faster by learning faster from more frequent iterations. Apple can only learn if their stuff works every year. Xiaomi can learn something every week.

[glassheart]

If a website makes an update, all users will get the new version as soon as caches expire. Something like an iOS update can also reach a large majority of users within weeks. Hardware that you've already shipped will remain in customer hands, obviously unchanged, yet still need to be supported through the warranty period and beyond. I don't understand why you would say that's nothing different. Your QA department would need to test software against many more versions of hardware, your software engineers would have to fix those bugs, and your after-sales support staff would have a harder time as well.

Also, it's not as if other companies ship exactly the same hardware throughout the lifecycle of a product. New suppliers are added, parts that have been identified as a return driver are reworked, and even major mid-life changes (consider the slim versions of each PlayStation as an example) happen as a matter of course.

[37prime]

Aren't we talking about hardware? It is a lot easier to patch software.

Hardware? Not the case. It is not practical to keep changing hardware when the production is larger than what Xiaomi is currently delivering.