Bitcoin is Technology based.  Because of it's dependence on electricity & connectivity, it cannot function in an offline world.  In other words, if a situation occurs that disrupts the internet or electricity, either on a global or local basis, Bitcoin will cease to function.  For example, suppose a country becomes isolated on the internet, either by it's own choice or by external edict.  Until enough nodes function in the isolated area, confirmation of transactions will not occur.  In addition, those that chose to use online wallets, may not be able to access their online wallet depending on where the wallet is located.

Bitcoin is a technology, and I don't think it was never meant to be used in physical form, like paper or coin.  Although, some people are trying to make bitcion more convenient, each individual should analyse the big picture.  Trying to turn Bitcion into a currency that can compete with the physical fiat type currencies is an interesting idea.  However, don't assume this has already happened, at least near term.  Please don't forget, Bitcoin is still an experiment in progress.  

As a side comment: Online wallets are convenient, but have some weakness - loss of access, dependence on a third-party for security, embezzlement - ie possibility of access by the online wallet provider.  It's my opinion, Bitcoin users needs to be aware of the potential problems when straying away from the official client.  Bitcoin is based on a technology, and that technology is implemented via the official bitcion client.  In other words, users of non-official clients, should accept the flaws and weakness of client and programmers who implemented the non-official client.  Yes, I would agree the full block chain(even a pruned chain) is inconvenient, but necessary from the view of the original Bitcoin Technology to avoid weakness.

Even the current system based on paper transaction system has issues in a disaster: DTCC is probably the bigest backend system most folks don't have a clue about, and it's having huge issues http://www.dtcc.com/downloads/legal/imp_notices/2012/dtcc/z0035.pdf

On a side note, normal banking too would be fucked by outages and offline. You would not even be able to withdraw your money from a bank without electricity. So what apply to bitcoin also apply to dollars and euros.

Given enough evolution of the technology, the stream of new Bitcoin transactions and blocks could easily be fed in through a satellite downlink - the same way XM Radio, DirecTV are downlinked.

This is no joke.  Satellite downlink is a simple one-way stream of huge amounts of data, and those streams can and do contain all kinds of things you would never expect.  If you've ever noticed "XM Traffic" on a car's nav system, you've seen one basic example.  If you've ever seen your satellite TV receiver update its own firmware over the air, you've seen another.  Same yet again if you order a pay-per-view movie and your receiver suddenly knows you paid for it the minute you do so, even though the receiver unit itself may not be connected to the internet.

If you've ever been the pilot of an airplane that supports XM's aviation downlink product, there's gobs of other information being downlinked through the very same satellite signal feeding tunes to your car on the highway - such as locations where lightning has struck recently, text-based weather/visibility condition reports at airports ("TAF's"), all of the doppler radar across the continent, and much more specific to the aviation industry.  All of this occupies a tiny slice of the total available bandwidth available in that satellite signal, and it's very profitable for the satellite companies to offer it, in terms of what they can charge for that feed. http://www.xmwxweather.com/ground/data-service-pricing.php

The Bitcoin block chain is no different.  With nothing more than a satellite feed, in theory you'd have everything you need to be a Bitcoin node, without the ability to send transactions.

Where there is a worldwide demand for a data stream, especially a relatively small data stream whose cost to acquire is nearly zero, the business case for a satellite feed is easy to make.  Bitcoin bandwidth requirements will grow over time of course, but just look at how many channels you can get over DirecTV today - easily into the gigabytes per second - Bitcoin transaction volume has a lot of catching up to do to fill it.  Meanwhile, you'll have demand for multiple providers of this feed, as people will cry "centralization" and worry about economic crippling if there's just one. (similar to GPS vs GLONASS etc).  So we could easily see "datacasts" of block chain data being packed into unused bandwidth of TV signals etc. and multiple independent ways to get block chain data off the air.  http://en.wikipedia.org/wiki/Datacasting

On the other side, sending individual outgoing transactions could be accomplished over regular radio waves, being short blasts of information that travel one-way on a best-effort basis no differently than a distress beacon.  Imagine you're a hut selling bananas in Africa and use solar to power your gear.  Doing a transaction might consist of nothing more than blasting it out on the open airwaves and then looking to see it relayed from the satellite signal within the next few minutes, trying again if not seen.  All of this on battery or solar or electromechanical power, no less.

I don't necessarily mean blasting the transaction back to the satellite (though it's certainly one of the good possibilities), but I am thinking more that you could blast a transaction using a low-cost transmitter with a range of a few kilometers, and expecting that somebody on the ground in range of your signal will either have internet access or the ability to forward the transaction somewhere else that does.  This would make the equipment needed to send new transactions simple and affordable, not to mention keeping it anonymous and open to anyone without needing a costly satellite "outbound data" plan.

In areas with sparse infrastructure (like Africa), highly directional antennas and installing them pointed at known receivers would easily extend that range into the tens of kilometers.  Better yet would be the definition of a "bitcoin radio protocol" that allowed a bitcoin relay network to form based on loose peer-to-multipeer radio broadcasting rather than an IP network, where transactions would hop over radio waves from node to node until reaching the internet at large.  This would be for propagating transactions produced locally and getting them on the wire, not so much for propagating the block chain.

While none of this is achievable for Bitcoin in 2012, if Bitcoin is going to take the world by storm, then being able to transact in this way eventually can be pretty much considered a given.  Hello Africa.

Casascius - what is the difference between what you propose to be done via satellite and a typical Internet via satellite service?  Why re-invent the wheel?  If someone has satellite service, why would they not opt for an internet connection instead of some weird listen-only connection?  Or are you saying this satellite listen-only service would be free, thus more accessible to more people?

Whether free or not, the main thing would be cost per user, exploiting the ability of the satellite to broadcast data at a nominal per-user cost.  If a million people get on a satellite and download the block chain using IP as the backbone (IP multicasting aside), the satellite needs one million times the bandwidth to send the packets to users, one at a time, as well as dealing with all of the overhead of acknowledgments, retries, etc.  But if the satellite simply streams the block chain as though it were a channel that everybody could tune to, the bandwidth effectiveness is maximized.  The stream would contain forward error correction and a small measure of scheduled repetition so that receivers who miss some of the transmission can fill it in.

I think we will see a more p2p-based internet in the future and more mesh-networks. If our smartphones for instance created a mesh-network automatically internet connection could be sustained with a few small solar panels and some smartphones in almost the worst of disasters.

Hologram BTC coins of various sorts could also work nicely in a dedicated community until connection was reestablished.

Actually, it can.  I've done it personally in a couple of different ways.  Cascious physical bitcoins is one obvious way, but I've done it by ad-hoc wireless.  The bitcoincard smartcard based hardware client works in much the same way, and limited transactions can occur in the absence of live connectivity.  The trick is that the sending client must have copies of the input transactions that it's using, must be able to send them to the receiving client, and the receiving client must have some method of checking those input transactions for credibility.  It's not as trustworthy as a pair of full clients with live internet and updated blockchains, but for most common interactions one might have in meatspace it's easily as trustworthy as paper cash, since cash can be counterfitted accurate enough to fool most inspections by the unaided human eye.

UHMMMMMMMMMMMMMMMMMMMMMM two-way sat phones cost no more than any regular cell phone, and alot of them offer free incoming calls with text messages costing mere cents.

Um, I'd like to see that service offer.  My experience with satphones is that they cost at least 5 times as much than a comparable domestic cell service.  Yes, I've looked.