Is a compression factor of 2 compared to a free codec worth the license trouble and the additional development?
Yes. This is why YouTube never implemented Theora, waiting until VP8 (which roughly compares to AVC baseline) before adding any free codecs.
What matters is that the existing standard doesn't empty the battery where a new codec wouldn't.
Why wouldn't the new codec be GPGPU-accelerated too?
Once a format is deemed "good enough" it can stick around for a long time.
True, if it is impractical to deploy a new codec in the field alongside the existing codecs, a first mover will win. This is why U.S. OTA digital television is stuck on DVD/SVCD era codecs, but some countries whose digital transition happened later use H.264.
Furthermore bandwidth prices have dropped through the floor in recent years
Long haul yes, last mile no. Satellite and cellular ISPs tend to charge on the order of $10 per GB. Even wired home ISPs such as Comcast and Verizon have been practicing "congestion by choice", refusing to peer with L3.
And, it becomes just more BSD code when the patent expires in... what, a decade?
A decade from now, most major web video streams will be in H.265 (HEVC), and H.266 will be the Next Big Thing(tm). By the time the patents on one codec have run out, bandwidth constraints cause providers of non-free media to switch to a new freshly patented codec. Users end up stuck on a treadmill, from H.261 to MPEG-1 to MPEG-2 to H.263 family (Sorenson Spark, DivX, Xvid) to H.264 (AVC) and so on.
If you want Android, but don't want Google Apps, you simply get a phone that is configured as such.
Which such phone, other than the Fire Phone by Amazon, is sold in the United States?
Stop whining that you bought a product and it is doing what it is designed to do.
We're whining that all products sold near us are designed to do something that we find undesirable.
A novel idea perhaps, but just maybe they should not try to push/throw everything into a webbrowser ?
In some cases, it's either deploy one JavaScript web app or deploy 15 native apps, one for each of 15 platforms. You can get the web app designed, implemented, tested, and deployed before you even become approved as a developer on half of those platforms.
Every successful person has had failures but repeated failure is no guarantee of eventual success.