To me:

Open Source Firmware signed by OS > Firmware blob signed by device manufacturer > Firmware blob hardcoded by device Manufacturer

The FSF treats hardcoded firmware blobs as "free" and updatable firmware blobs as nonfree despite there not being a big difference between them in practice. And practical differences like being able to fix security issues benefits users.

> And practical differences like being able to fix security issues benefits users.

More often than not these updates are not actually benefits to the users.

Can you provide such an example? Because of bugs in the new version? A lot of the time old versions can still be loaded.

Just in the last couple of days:

https://news.ycombinator.com/item?id=45568700

There are by now thousands of examples of this, I wonder why you would ask for an example, this is about as uncontroversial as the sun going up tomorrow.

The number of times that Linux distros, free software, has made my computer unusable, requiring me to manually fix it, is uncountable. Bugs from OS updates is still entirely possible even without updating firmware blobs.

To a driver whose car just died on I 405 in heavy traffic that is a distinction without a difference.

> that is a distinction without a difference

Yes, agreed, the entire idea of OTA updates to cars has a lot of bad consequences. No matter the license.

So when we analyze the entire position of the FSF, all three cases listed above, I don't think you actually agree with them. FSF isn't against updating firmware on the fly, they just want certain things when it happens. And those things won't improve safety.

Agreed.

But: Stallman & the FSF are idealists, they are not in a position to argue for practical measures, by definition they have to take a relatively extremist position and dig in. If they didn't do that they would get nowhere. So I understand why they're doing it but I do not think their stance is overly practical or workable. But everybody will dilute that stance in their own particular way and so the net effect is potentially positive. If they would take a much milder stance then that positive effect would be diminished.

I would say that exempting unchangeable blobs is a big concession all by itself and the idealist position is that roms with programs need to be open source too.

Yes, I guess they gave up on that one realizing that it just isn't going to happen.

I don't even mind the ROMS, I mind the ROMS that I can not read out myself.

Thank you for clearing that up, I clearly misread that completely.

I initially made the same misread that you did...

The OP's point is, having the firmware permanently burnt-in on a ROM chip vs loaded as a binary blob via a driver doesn't change the "non-free"-ness of the firmware itself.

So opting for hardware which has a "fully-open-source" driver, but runs a binary blob encoded into the hardware, doesn't make the system fully open.

It's a take for a more Free system, not for accepting binary blobs.

(Or I guess for acknowledging that if you're willing to allow binary blobs stored in hardware, then dynamically-loaded binary blobs doesn't change the "free"-ness.)

That's not even close to what they said.

They're saying approval of any who-knows-what code shouldn't be decided based on how it's loaded.