Here on earth our nuclear power comes from fission reactors producing 500-3000 MWt of heat which we in turn attach to a glorified thermal power plant consisting of water moving in a closed loop through pipes and turbines not fundamentally very different than a steam engine. This nuclear power is instead the radioactive decay of a small radioisotope, emitting a mere 500Wt (but with no way to turn it off because no fission is involved) which we in turn attach to a relatively unconventional thermoelectric generator which doesn't require the giant size or any moving parts (except electrons) which a terrestrial thermal power power plant employs.
In both cases you need a heat source (the nuclear fuel, probably ~500K) and a heat sink (either a large body of water at 300K or the outside of the spacecraft which is a cold ~100K because it radiates any heat into the vacuum of space) and generate electricity from the difference.
The utilities are sticking their heads in the sand and trying to pretend that technology won't move forward. In some places they are trying to add an interconnect fee for those with solar panels that's as large as my electricity bill. They also are requiring solar panel inverters to stop working entirely when the grid goes down, instead of just providing power for the house and still leaving the grid upstream unenergized. All this, and the price of electricity keeps going up. And they expect people won't move forward with batteries as technology improves?
Disconnecting from the grid entirely is large investment: people need a large solar array, several days worth of batteries, and probably smart appliances (mainly air conditioners and refrigerators). Or the utilities can make money helping to create a lower-investment intermediate option: staying connected to the grid with a smaller solar array and half a day worth of batteries which both help the utility with load balancing and can keep the house powered when the grid goes down. If they do this right, they will be able to remotely control when the system is storing energy or sending it to the grid, which probably means it's in their best interest if they write the software and maybe even make and sell (and install?) the hardware.
Plus, they can provide monitoring services and, if they want to really diversify, insurance services or financing options. Otherwise, as more people abandon the grid, it will become more expensive per person to maintain it, creating a downward spiral of grid usage.
The evidence claims that so far, there hasn't been an increase in monetary cost of natural disasters relative to GDP. I'll let other, more informed people tackle this factual issue.* My problem is purely based on faulty logic; at the end of the article, the author extrapolates that this trend of disaster damage being correlated with and caused by increases in GDP will continue indefinitely. But the only evidence cited for the conclusion that climate change won't ever cause increased natural disasters actually says that US tropical cyclones won't significantly increase in frequency and severity for several decades; I found nothing about winter storms/polar vortex, crop loss due to drought, sea level rise, etc. and I'm not even sure how accurately you can extrapolate to tropical cycles in other places, not to mention many of us hope to still be around in several decades. I appreciate that Nate Silver is a great statistician, but this is going to go downhill really quick if the conclusions of articles posted on his site are only tangentially related to the actual statistics.
*The other disappointing thing is that the author has claimed this before, has been refuted, and hasn't changed his argument even so much as to mention the points made by various people who had rebuttals.
Other people have addressed several of your issues, so I'm just going to look at the claim that driving uphill empties those batteries in very little time.
As far as I can tell, the steepest drive in the United States (from the beach to the top of Mauna Kea, HI -- not entirely paved) is approximately 13800ft / 4200 m above sea level. Since the mass of a Tesla Model S is 2100 kg, this would consume 25 kWh of energy (30 kWh if you loaded it with 300 kg of people and stuff and factor in a 93% motor efficiency). The energy stored in a full Tesla battery is either 60 or 85 kWh, depending on the model. The drive from Hilo is 43 mi / 71 km (or Kona is 64 mi / 103 km), which over flat terrain would consume 15 (or 20) kWh, for a total of no more than 50 kWh. Thus, you could easily do the drive in a charged 60 kWh Model S. And, the drive back would be entirely free because you just brake all the way down -- just don't charge it at the top unless you want to burn out your brakes.
In theory, the potential energy of a Model S (+300 kg) on top of Mount Everest is 59 kWh, so I don't recommend that with the 60 kWh battery, but then I suspect you'd have issues trying to drive any car to the top of Mount Everest.
How are you going to get the groceries back home, make 10 trips? I think I'd prefer to drive.
100m?! Unless you're on crutches, 100m to and from the grocery store is no trouble at all. I used to walk to the grocery store when it was over 300m away and I just carried a bag in each hand and (if necessary) slung a bag over each shoulder. If you prefer, you can bring a backpack. If it's raining, carry one fewer bag so you have a hand for an umbrella. Seriously, unless you're at Costco buying a TV and four cases of beer for a superbowl party, walking a limited distance to do your shopping most weeks isn't hard.
Even if you live 3 miles from your grocery store, all you need is a backpack and maybe a basket or two on your bike and you are set. If you have kids, hook up the kiddie trailer and fill it with groceries too. It's not that hard.
Believe it or not, most of the time most bicyclists don't like having to share the road with cars either (they belch exhaust right in front of you while you're trying to get as much oxygen as possible, they like to cut you off, they like to swerve into bike lanes without looking before turning, they drive at the speed limit even when it's not safe to do so due to poor visibility, etc.). If a bike is "in your way" chances are it's the only reasonable way for them to get to/from work while on a bike.
Sidewalks (in places where it's even legal to bike on sidewalks) often have low hanging branches, disappear without warning (either into a curb, 100-year old tree, or someone's yard) and you have to go a lot slower because if you're doing 20mph on a sidewalk it's not going to be long before you're going to run into some car backing out of a driveway that isn't looking that far for you.
Which is why there should be bike lanes on more roads. But even then, if you encounter a roundabout or want to make a left turn, you have to deal with cars, and cars have to deal with bikes. As it is, I go an extra 2/3 of a mile out of my way each trip to avoid cars as much as possible.
The fiber provider in my area (Canberra, Australia) has options for n GB(8am-2am) + n GB(2am-8am) where n is 20, 100, or 500. It's $100/month for the 500GB+500GB option at 100(down)/40(up)Mbps. Dropping down to 100GB+100GB at 12/1Mbps costs $60/month, which is what I was paying in Berkeley for Sonic DSL two months ago, and is pretty much the speed I was getting. Once you hit the cap, you're throttled to 256kbps both ways.
Seems reasonable to me. The more I use, the higher percentage of their "tubes" I'm using and the more they need to build.
I got a 500GB/4GB one about three years ago. It's definitely faster than my previous (250GB/5400 rpm) drive, but it's also a lot more power hungry and seems to rarely spin down. SSDs, on the other hand, have much lower power consumption and are way faster. If your computer has USB3, you're definitely better off getting a small SSD and a 2TB 2.5" USB3 external drive.
PS. My first one died within a couple months, and this one's SMART says it's been failing for nearly 7 months now, but I'm stuck waiting until it shows a clear symptom or fails Seagate's own utility before I can RMA it.
Happiness is twin floppies.