The big difference between corporations and government is accountability.

Corporations nowadays are semi-directly accountable to a comparatively small group of stakeholders, and they are basically designed to be sociopaths. All I can do is vote-with-my-wallet, the feasibility/effectiveness of which is highly dependent on circumstances and usually just an after-the-fact measure.
Governments in a democratic society are designed to be fully accountable to the public, e.g. you and me. There is absolutely no problem in letting such an entity become really big, as long as it can do nothing else but represent and serve the people. Corruption is always a problem, so anti-corruption efforts should be a major goal.

Now in the particular case of the USA: I don't think you can honestly claim that this government democratically represents the interests of the population of the USA. Fix that, don't just treat the symptoms by aiming for a smaller yet still corrupt government, while at the same time (inevitably by the efforts of current libertarians) giving more power to corporations who are guaranteed to behave like sociopaths.

Limiting organisations to 100 members is interesting, but they would form coalitions of 100 members (each a 100-member organisation) and meta-coalitions and so forth (even if it's just inofficial). Not far from divisions and subsidiaries today.
Instead, I think accountability is more worthwhile (e.g. employee-owned corps)... or its little cousin, transparency (e.g. all board meetings should be broadcasted to each employee or even the public).

If by "developing its capabilities" you mean "analysing, understanding and applying NASA knowledge from the last 5 decades" to which they have full access then yes, they did that at some point and are still doing it. However, I'd be very surprised if their own research added even close to 1% to the heap. Just look at the outright silly disparity in amount, scale and scope of experiments, the size of the funding and R&D staff, etc. between the two.

They are basically a private extension of NASA with a significantly less risk averse decision making process, but also much less accountability. Not that I have anything against that, I think SpaceX is awesome, but I also do think that Tyson is mostly right.

Who tells the machine to take a sunset? Who enables it to choose? The artist.
This robot and any software picking & repainting google images is exactly as intelligent as the painters brush, just a bit more complex, and has no more self-initiative or creativity than a piece of wood.
Taking a picture with an expensive DSLR doesn't make the camera the artist, and mounting it on a self-driving car that randomly takes snapshots still gives all the credit to the person that built this.
And photographing the Mona Lisa with a lens that introduces imperfections (that's what is done here) is of questionable artistic value whether done by a human with a camera or a human with a robot..

Don't get me wrong, I'm often annoyed by those interpret-the-world-into-three-strokes types (likewise, literature teachers who "know" that the author meant to convey this or that theory), but pushing more of the manual work into the tool (brush or camera or robot) doesn't make it an artist. For comparison, a machine will never own a copyright under our current concept of the idea, only the user/initiator/owner/... of the machine will, no matter how automated it is.

In summary, both the guy spraying paint from his ass and the one building and programming a highly complex robot to paint pictures produce art, but it's not the robot or the rectal muscles who take the responsibility and are called artists.

Who cares about a post-apocalypse tribal society on a pre-modern tech stage trying to restore mankinds' knowledge?
Give them 10k more years and they'll manage to do it with femtocell lasers just fine. Or 50k years, it really doesn't matter, it simply shrinks compared to the idea that some cataclysm just wiped out all books accessible in the world, all professional knowledge, reading skills, parents-teaching-offspring, dozens of other information carrying media types (respectively it's usage knowledge) that would be around anyway etc. which could allow them to get up and running more quickly... but somehow left a few humans alive.

This storage type is not meant for a post-apocalypse tribal society restoring mankinds' knowledge. But some of us would be happy if the now often unreadable magnetic records from 70 years ago would have been stored on something more durable.

We have no problem paying for what we use. But we dont want to pay for the things we dont use. Like the wars, the spying, the surveillance. And the things that we do use, we want provided in a competitive market-place where abusive unresponsive or otherwise problematic suppliers cannot simply continue to bill us as much as they wish and use it for whatever they want!

I'll hijack this here, because that's an important point. One of the strongest correlators of crime is the (inverse) quality of public education [0]. Paying a small share for the education of your neighbours kids, even if you don't have any on your own, means paying for things you don't use. Yet the reduced stealing, robbing and killing 10 years down the road will benefit you personally.
If everyone has a choice, many will not pay. If you don't pay, either you expect others to and are leeching off of them, or nobody does and you don't have whatever benefits come from highly scaled solidarity/collectivism. You're losing something valuable.

IMO education and health care clearly fall under this category (and I claim most people who lived in a country with universal public health care would agree), along with the obvious infrastructural stuff. The spying, offensive wars, etc. definitely don't. As mentioned, the modern state can be used for both good and bad, and instead of removing all of it (which leaves you in a worse situation, even though Libertarians call it paradise), why not fix it by returning your government to its intended role as a servant to the people? That's the point of democracy, and the symptoms you are fighting are results non-democratic seizure of power by special interests.
You can say that there will always be corruption, or that the powers that be will not allow change, but the first can be mitigated in various ways, and if the second holds, your fight would be futile to begin with.

[0] Before anyone brings it up (aka inb4 guns), as a simple example for why gun control doesn't matter that much: check Honduras (high gun proliferation) and Jamaica (very strict gun laws), versus Switzerland (high gun proliferation) and the UK (very strict gun laws). The first two have high crime rates (and below average public education) while the second two rank lowly in world crime statistics (and have what is considered high quality public education). The main effect of gun policy is the percentage of gun related deaths and the location of crimes.

And in the system I advocate, you could say that, while in the system you defend, you cannot. Or you can, but it wont fly, of course. Find a way to refuse (difficult in europe, where the money is nearly always taxed before you get it) and you will eventually see why we say the power of the state grows out of the barrel of the gun.

If it's completely optional, in the (for the lack of a better expression) "heat of the opportunity" 500 of the 1000 will refuse because they really need the money for something specific right now, and they already helped some guy last month, and, and ....
Another 300 will refuse out of greed, another 150 will point out that they have a choice, and this doesn't match their ideology or they don't feel compelled enough morally.
So my share grows twenty-fold, can't give that much because I also divert a share of my income to maintaining our roads, electricity infrastructure, military, etc. And for some it snowballs into reducing their willingness to pledge help because it leaves them with too little disposable income. You go hungry.
Solidarity requires broad coverage to work. Voluntary charities don't cut it, at least there is no example of sufficient equivalency out there, e.g. funding a national education infrastructure or healthcare program. Also, the overhead of your proposal leads to higher bureaucracy if it is to handle a similar volume of money assignment.

But why should I try to refuse paying taxes? Excuse the crass example: try to enslave people, which might be good for your personal profit if you could keep it that way (and "some" people would do so if it were optional)... you'll still get stopped at gunpoint, presumably by the government.
I think both enforcing taxes and personal freedom is justified ethically and in terms of the net benefits, and accordingly required/guaranteed by law.

One paradox you face is that while a relatively wealthy nation with a culture that values work and production can maintain a welfare state and pay for it with taxes without destroying your productivity overnight, you still sap the very qualities that make this possible in the longer term. Which is to say, over time doing this makes you a less wealthy nation with a culture that places less value on productive work.

I require proof of that. Clearly you can't give it for now, but I'd like to point out that this is a problem of blame attribution: you see the cracks (which are always there in some form or another) and attribute them to a certain ideology you're compelled to oppose. I'm guilty of this too, it's human nature, but still not a good argument when reasoning about the merits of a system.
As for your proposed alternative, I see how Libertarian state could work in theory, but not under the requirement that it matches a metric like the GNH (yes, I'm actually serious about that) of current states. As mentioned, IMO absolutely not worth aiming for.

With regards to the "USA protecting the world", I wouldn't say the out-of-bounds spending on your military-industrial complex is benefitial to Europe's safety. Apart from that, Europe is easily second, so there's only one entity in the world we'd have to watch out for ... ;)

On Sweden: they have utterly failed in their immigration/integration policy. You can blame it on the conservatives if you like, but claiming to see the writing on the wall for their whole system is far fetched. See blame attribution above. And you'd have to elaborate on the Swedish educational system being libertarian in nature, because most Americans so far have called its current form socialist (or communist, you get my point).

Not enough. I had a very bad year, broke my leg and been out of work for months. Got me an the missus and 7 kids. A small portion aint gonna do it buddy. Now where's our food?

I'm still only going to give you a small portion of my income. 0.1% should suffice. And my 1000 friends here will too, so you'll be fully funded instantly - such is the power of collectivism. And once you don't need it anymore you'll work with net gain by yourself again, because we're living in a capitalistic economy. And society won't have a valuable contributor starved, plus one of your 7 might be the next Einstein, so it'd be a shame if he/she missed out on a complete high quality education because it wasn't free or very cheap.

Surprisingly I hear from some US citizens that this wouldn't work, which doesn't make sense since most of Europe is proof that it does.

I know that you don't want to forbid the farmer from helping (PoC), but the case for forcing him to help (within reasonable bounds obviously, though that's another debate) is far stronger IMO.
I don't understand where you see the indications that this is the path to failure, and such reasoning seems irrational to me.
Why should it lead to breakdown and not prosperity, as shown by societies that apply on a large scale (some of) what we already practise within a family.
Likewise, Marx' claims that all capitalism inevitably leads to breakdown and then communism still lack any indication of proof, and he made a highly elaborate case.

Let me ask you in return: how would a society as proposed (how I understood it) by the Libertarians not devolve into right-of-strength anarcho-capitalism? I don't see such a place as desirable to live in.

I agree that world hunger is technically a distribution problem, caused by a political problem. So how to fix this and similar issues? Obviously "fix" the entity responsible for politics, i.e. the government, only that our ideas of "fixing" are highly divergent. I've made a proposal here.
The key idea is to make the government fully accountable to the people, not remove most of it. In this regard, I couldn't have written mrthoughtful's comment below better myself. What you call "the modern state" in your reply is a highly useful concept, similar to money or mass media (all of which can be used for bad).

His definitions are not faulty, they just don't match yours. The farmer can and should be forced to give a small portion of his sellable produce to fulfill someone elses right not go hungry.

I assume you live in the USA, a country that is currently way closer to your ideals than his. He's stated to be living in Germany, which is organised in a manner closer to the philosophy he describes. You seem to hate the way your society is organised, he seems quite happy.

The liberty to do what you want with 100% of your work without any obligations to society is... well, it's a goal some may choose to pursue, but why should we put it above things that are more important in the real world?

Disclaimer: as a western European with a center-right political attitude, I don't agree with all of what he said in his original comment, but the first paragraph and the fact that taxation is a Good Thing (TM) is pretty much common sense.

The US badly needs something like the Swiss referendum and initiative, where the population can overturn pretty much any government decision, although the executive ones are harder to get at because it goes indirectly.
I think that'd shut up the "sole arbiter" complaints (disclosure: I'm Swiss). And if you then also increase transparency and accountability of the government agencies, the whole Libertarian Party starts to look as silly as it would if we had one over here.

Give me a dumb framebuffer _with_ network transparency and I'll happily shut up. If they finalise the design and only tack on networking afterwards, it'll be probably less efficient than it could have been.

I for one want them to prioritise it during the earlies phases. They are telling me (indirectly) that they probably won't do it at all and leave it to someone else. That's fine, but also means Wayland is designed to not be used by me and people like me. And when distros start pushing it as default, there will be a big ruckus... I'd rather them to be sensible and support network transparency at least as well as X11 does, and that it "just works" the first time I have a Wayland server in front of me, for all applications that don't have Wayland releases (and obviously the new ones too).

I think the author of the post you replied to originally didn't actually intend to be taken seriously. As such, I'm assuming that you're simply playing along and not really expecting there to be a project in existence such as the one he mentioned (except maybe for DVL).
That's why I left this aside and focused on what I percieved to be the crucial part of your comment ("Is that a tragedy of the Microsoft-ocracy keeping the world closed for users?"), where the reasoning is IMO lacking.

But addressing that other part of your point: Commercial FOSS and proprietary software have similar user friendliness. Likewise, hobbyist FOSS and closed share/freeware apps both often a hassle. Developer attitudes are about the same (my view after having worked 2 jobs as proprietary and now commercial FOSS developer), it depends on whether you have first line support and sales in between. Being allowed to look inside is certainly no impediment, the problem is closed-source copyright licenses: can't publish any modifications, even though you can make them privately while infringing on the EULA (which is legally shaky anyway) since nobody knows at that point.
If we're going by examples, I wouldn't call Ubuntu less user friendly than Windows 8, and the random hobbled together CMS on SourceForge is no more a pain in the ass than the proprietary Contrexx* (terrible experience it was... of course, there are good both FOSS and proprietary CMSs). I agree that some projects are annoying to get running, for me there are clearly more on the FOSS side, but then again I've tried out _way_ more of those, since I don't go shopping for dozens of randomly picked software packages and closed freeware is somewhat scarce on my system of choice.

But returning: as already said, even though only few choose to disassemble their Ford, the important part is that it stays _not_ forbidden, independent of the users inclination to actually exercise the option. In the software world on the other hand, we sadly have that situation, which is where I'm having trouble with the analogy.

* a rare case of proprietary "but the source is visible" software, not Free, nor anywhere near OSI approved, but they keep calling their license "Open Source"

This is a fairly naive view.

If we make hurting people and destroying their property illegal, we can abolish every and all traffic laws. After all, if people obey the law and simply don't hurt anybody, there will be no deaths on the road even without speed limits, regulations on right of way, car safety standards, etc, because the "no killing allowed" law will somehow magically trigger and make the other laws redundant.

I don't get that car analogy.There is a difference between technically being able to acquire all tools and knowhow to replace any particular part on my car (=FOSS)... and being strictly forbidden by the manufacturer to open the hood, being threatened with legal action if I pay some independent expert to peek inside and being told it's all "for my best" (=proprietary software).

I'm allowed to disassemble my Ford (though few choose to), but not my copy of Photoshop.

This particular terminology is System/Kernel, and a not at all exhaustive overview would be:
GNU/Linux (Debian, Ubuntu, ...)
GNU/Hurd (Debian has one such distribution, Arch has another called "Arch Hurd")
GNU/kBSD (Debian has one, [0])
GNU/ON (Illumian, a Debian+Solaris kernel distro)

GNU/BSD or GNU/Solaris would be somewhat more like GNU/GNU, which doesn't make a lot of sense.

[0] They call it "Debian GNU/kFreeBSD", as it uses the FreeBSD kernel, which isn't actually called "kBSD" or "kFreeBSD" but has no distinct name, as it's not Mach anymore, but simply "the kernel of FreeBSD"

"Just" a microkernel is fairly straightforward to build if you have a bit of experience. I could make a simple one in two afternoons.

Microkernel OSs with a very limited feature set and somewhat OK-ish performance are feasible, as shown by e.g. velOSity and embedded realtime systems like QNX.

Microkernel OSs that can be used for general purpose computing and offer a nonvanishing fraction of the performance of Linux, XNU or NT are ... well, the development times of the L4 family, Hurd (OK, this one has other roadblocks), Sawmill etc. show how hard it is.
There have been quite a bunch of microkernel research projects (by at least IBM and Microsoft for what I know, plus the usual bunch of universities), and there has been renewed interest lately. In the future it might come down to "your hardware is fast enough that a performance hit of, say, 50% doesn't matter, and the added stability, security and MUCH simpler life for developers that interact with the lower parts of the system are worth it".