One of the big issues of this platform is how things are being changed every single week. It's hard to keep up with.
Also, until the platform is allowed to settle, there isn't enough statistical data to justify any of the changes. The changes seem to be made on gut feeling, rather than with taking a month's data, analysing it and then making changes. There have been so many higgledly piggledly changes that it's now impossible to do any analysis on what works and what doesn't.
Indeed, the slower approach you describe here is where real engineering could be very useful, albeit slower, and investors don't like slower. We're living in an ever increasing atmosphere of short attention-span, microwave oven addicts. This is where the balance I spoke of earlier comes into play.
It is also important to realize how important a factor time is. If you wait for a product to reach perfection in its' evolution before you release it to production, it may never be adopted b/c someone else gets to the market sooner.