My only problem with underscores is that when I read them aloud in my head, I hear a longer pause than for a hyphen, so code with lots of underscores takes longer to read, or at least feels like it does.
I can just imagine receiving a link like this while I'm on a spotty WLAN, trying to load it, it not loading properly, refreshing, and having it be "expired".
* Keep the coffee machine in the bedroom. Before going to bed, load it up with water and coffee put the alarm clock next to it. Also have a cup and some milk next to it. Next morning, when the clock goes off, you're going to have to walk over there anyway so you might as well turn on the coffee machine. Now you can go back to dozing off but after a few minutes, the scent and prospect of the readily available coffee should make getting up much easier, plus you can sip the coffee in bed.
* Count to ten, and get up on ten.
* Take melatonin the night before to go to sleep on time in the first place.
* Log your sleeping hours to get an idea of what irregular or overlong sleep is costing you in terms of time.
I find this USB stick distasteful. It's not mechanical, all the gears do nothing. The designer took parts that had a precise function and rearranged them into something useless for appearance's sake.
This is probably a weird reaction, but maybe other hackers felt the same way.
Even if flying cars were cheap and easy to make, would we want them? They're cool and amazing, but in real life they might be impractical.
So generally when talking about the future, maybe we predict inventions that make for cool easily understandable stories, instead useful or plausible inventions. That's why flying cars were predicted but the internet wasn't.
If you're not even going to be alive when the time of your prediction comes, why would you bother to think hard about it and make an odd, abstract, unfashionable prediction instead of a cool story about flying cars?