Is it actually "neglected" if the existing service works reliably under a continually decreasing load [0]?

Surely the actual time in transit makes up the overwhelming majority of time of DVD service and mail is generally a once per day delivery.

Not much point in optimizing what isn't broken in the electronic portion of that operation.

0: http://gigaom.com/2013/10/21/when-will-netflix-kill-its-dvd-...

From another point of view, one could say it's already been fine-tuned and is running quite well as is. In that case, why bother messing with it?

On some level, this is the profitability dream. Engineer something that doesn't cost a lot to run, doesn't need to be changed, and makes a ton of money. Engineers don't like to hear it, because it means less demand for our services, but it's what's best for business.

You can't always drive increased revenue through increased R&D. At this point, substantially everyone who wants a DVD delivery service uses Netflix. As long as someone doesn't come out with a service that makes it significantly easier to rent DVDs over the internet, and as long as attrition caused by the software is very low, any more money spent on Netflix's DVD service is throwing bad money after good.