"all over the place", come on.. Those were just my 2 cents.
I may appreciate the efforts of the Hurd developers. But, as BO$$ meant, an operating system which, on modern computers, won't work properly (no USB=No keyboard, mouse or thumbdrives) or install (no SATA= no hard disk install, not even booting from a modern DVD drive) is useless. And spending 23 years on a purposeless software is throwing your life away.
It looks like a group of engineers who share the dream of developing the most advanced steam engine in the world start working, their idea does not attract many people but they go straight on with it anyway, and after a couple of centuries they come out with a barely working steamer which "with still some, uh, loads of adjustments, fixing and developing it might work on some 1800's coaches" while the world all around them is traveling on hybrid-powered cars, jet airplanes, electric hi-speed trains and coming-soon maglev tracks.
Nope, this analogy is wrong.
Contrary to the comparison of steam-machines and hybrid-cars, high-speed-trains and coming-up maglev-tracks, the requirements for a Kernel haven't changed that rapidly over the years.
Granted, the Hurd needs more suppport for standards like SATA and USB, no question, but unless you are a kid like BO$$, focusing on the *direct* practicability of a Kernel, the advantages of the Hurd are clear:
BO$$ doesn't understand what translators are, so his points are ultimatively rendered invalid, as he does not know how the Hurd works in any way. He might be only able to think in the scope of a person just interested to play games, use proprietary software and benchmarking his system, not taking in regard the Hurd-codebase is not as well-optimized as the Linux's one.
The reason it takes so long for the Hurd-developers to implement USB, SATA and all this other fancy stuff we seemingly cannot live without today is the fact, that they attempt to implement it in a way it suits the design-specification.
No matter how well the Linux-Kernel works, it suffers from many horrible and lousy design-decisions which hinder the actual implementation of new features and standards, which are in many ways completely different from past implementations of old standards.
Aiming to be compatible with many more situations than what the Linux-Kernel could ever offer, we now may have a quite featureless Hurd, but hell, considering what has been implemented, it bloody works without lousy workarounds (cf. FUSE, which is an abomination to human kind).
You might ask: Why don't they start implementing the new features now? I answer: Why should they do it now?
No one really urges them to do it. If you have read into the topic, you would understand, that GNU-developers never really see the urge to follow corporate-demands (unless corporations and individuals pay them), but to design something which in itself is not faulty and prone to implementations of a rushed kind.
Run the Hurd in QEMU and use it for a while; you will eventually see what I mean, even if you might not agree with or understand what I mean presently.