There's really no substitute for electricity when it comes to personal communication, but it doesn't take very much. A few watts into a VHF or UHF radio can transmit throughout a neighborhood; into an HF radio, a county, state, or more. (Long distance commercial communication was done with heliographs (mirrors) on sunny days, and signal fires at night, but that takes a vast investment in labor and training.) Enough LED lighting to, say, read poetry or play cards, takes a fraction of a watt, and can be provided from a modest battery charged from the sun during the day, and is much cheaper than candle or oil lamp light... once you've got it built.

Years of reading JMG has persuaded me that lengthy slow stepwise decline is a more likely scenario than a fast crash. That said, fast crash cannot be entirely discounted as so many vulnerabilities seem to be present in modern industrial civilization. I personally do not expect it will be likely that I will be significantly impacted by either since I have attained the age to be largely parasitic on the younger generations and hence will not be around long enough.

That said, I find the fast crash scenarios to be more interesting as thought experiments. Over the years I have given a fair amount of thought to how one might prepare should they choose to do so. I think my best insight is that it is not useful to be mentally invested in any particular scenario as you very likely will be wrong. What is useful is to identify commonalities of end result. All the end results that are deemed very unlikely to be survivable are the ones to not worry about at all as by your assessment they are not survivable. All of the end results that are potentially survivable, regardless of the specific scenario that led to that result, are the ones to consider preparatory action if one is so inclined.

For me personally the worst end result to consider survivable would feature little or no grid electricity but society and commerce functioning to some degree on a very localized level. This brings to mind JMG's concept of salvage economy. In any localized society/economy survival is enhanced on an individual level by one's usefulness to the community at large. In preparing for a salvage economy it is worthwhile to prep in a manner that leverages one's personal skill set, whatever it might be, in the context of a salvage economy. In economics there is the concept of "stranded asset". This can be defined as an asset which in different circumstances has value but due to it's current location or circumstance has little or no value. A good example would be associated gas at an oil wellhead where there is no infrastructure to take the gas thus it is flared.

Finally getting to the original topic of this thread, an example of a stranded asset in a no grid electricity scenario is the investment in electrical wiring and associated fixtures in every home. Originally costing several thousand $ it is rendered stranded by lack of electricity. As Lathechuck very correctly noted earlier in this thread it takes very little electricity to provide a useful amount of lighting by way of LED lamps. Except for the LED lamps themselves and a few watts of solar panel everything else to make lighting possible is available as stranded assets. Initially 12V car batteries will be widely available.

A longer term solution for the battery needs are the ubiquitous Li-ion batteries in so many consumer items. These would require a bit more advance material preparation and associated expertise to provide necessary charge/discharge control as well as stepup or stepdown switching voltage regulation to match the design voltage range of the LED lamps. The LED lamps themselves could be a viable cottage industry were the components at hand and can yield much higher efficiencies than any of the commercial product I have examined which are largely crap.

Necessary control circuitry is very simple, as little as one transistor and one resistor though two of each yields much tighter current regulation. With surface mount construction simple circuits such as these can be fabricated with very few resources provided you have them. Reflowing can be done in a frypan if better equipment is unavailable. The lamp screw bases can be salvaged from compact fluorescent lamps. Stockpiled in quantity, component cost per 12V lamp utilizing 4 surface mount LEDs and capable of setting power up to 1.5 watts could be on the order of 25 cents per lamp. Efficiency can be better than any commercial product I have seen in 12V LED lamps.

So yes, I think low voltage DC will have a place in the future using the same household wiring and fixtures originally installed for grid supplied AC.

While you can actually design a simple, safe, power conditioning circuit to run LEDs from 12V DC, there are a couple of subtleties that come to mind. First, what are the tolerances on your "12V" power? Automotive "12V" power is actually 12.5 from the battery when the motor isn't running, but regulated between 13.4 and 14.5 when the motor is running. Solar panel voltage varies over an even wider range: a 100W panel rated for a 12V system might put out 18V (or more, depending on the load) in full sun. A simple circuit that works fine on 12V might fail catastrophically when hit with 18V, unless it's designed for that level of input voltage.

Second, a single-transistor "linear" regulator that is reasonably efficient with 12V input will be wasting at least 33% of the input energy when fed with 18V. On the other hand, a more complex "switching" regulator circuit can take quick "sips" of 18V, or longer swigs at 12V, and smooth out the flow to get the proper voltage and current to the load. (In fact, it could be designed to accept 100V.)

Third, just as you need to consider over-voltage conditions, you might want to consider under-voltage. Maybe you have solar panels that can put out 100W at 18V on a sunny day, but only 3 W at 9V on a cloudy day. If you'd rather have a 1W radio all by itself than nothing at all, a power-controller that can increase the voltage would come in handy. A buck-boost switching regulator can draw more current than the load requires, though at a lower voltage, and step it up to the required voltage (with less current than it started with, but enough for the load). (I don't know whether "3W at 9V" is a realistic point on the power vs irradiance curve, since it's so far below nominal that my sources don't tabulate it. In general, available current falls a lot faster than available voltage, as the light dims.)

Anything more than a 78xx voltage regulator and a few caps is above my pay grade.