Since usually you can always add open parts to a system, the question is not if there are open parts, but if there are (substantial) closed parts.

> Linux with a video card.

Assuming you are speaking about a recent video card from AMD or Nvidia, then that system has substantial closed parts, which are unreviewable and unmodifiable by the user and/or owner of the device.

He means the binary blobs.

Ideological and rational are not in opposition.

Also, he doesn't think all software is good just because it's foss; it can be bad for other reasons. But all non-foss software is bad, because it violates the rights of users.

I don't see the problem with being this consistent. If I say that all non-consensual sex is bad, am I an irrational ideologue too?

This. I think when you have one lens through which you view the world, as opposed to having multiple lenses to pick from, which you swap out to apply the most beneficial lens from which to view the problem .. then this is no different than using the one hammer you have because all problems look like a nail.

I do think there are some software that in reality, for most people (including large swathes of non-technical people), work better in a for-profit model (which is usually closed source).

But there are some things that work better as open source.

I think it really depends on the fundamental problem (i.e. ends) the software (i.e. means) is trying to accomplish.

I personally do not think you can categorically say one is better than the other, because it really depends on the issue.

Why isn't there an open source Google search, that works better than Google?

If we can open source our encryption algorithms because that ostensibly makes it more secure overall, why can't we just open source all our algorithms for spam filtering, especially to the spammers themselves?

If you look at all the most vibrant communities online, are they closed source or open source? (Twitter, Facebook, Reddit, HN itself) ... vs. say, mailing lists, usenet, etc.