Complaining that the GPU isn't open is one thing and I totally understand that, saying it sucks is completely different to me. Saying it sucks implies that there is something wrong with the hardware and/or its drivers. It would be similar to claiming that a Macbook Pro sucks because you can't swap the battery yourself.
Now, if someone can show where the binary file has a flaw in it so that it could be said to be defective in some way, then sure, let the accusations of suckage begin. Especially if the flaw prevents usage of the device as described and it remains closed so the community cannot fix it. But then if they don't fix it themselves nor allow the community access to fix it then the platform will die anyway.
So, back to my comment, I don't understand the point of the article and others like it.
Similarly, the Raspberry Pi was advertised as a computer you can customise and play with. It was built for educational use, and yet part of it is hidden away from study. This is annoying some people who expected more than they reasonably should have.
There's been no indication that the hardware itself, the chips on the board, nor the driver do not PERFORM as expected.
If the Macbook Pro was advertised in such a way that was not true in the end, yes indeed that "situation" would suck, but the computer itself would still be a good computer.
I think it's pretty clear. What broadcom released was a binary file and and code to interface with it. It's not that the driver functionality is any different than what's already available it's that all the important functionality is in the binary file that was provided without any source code.
They made a big deal about the first open source ARM gpu driver but, for the most part, it's still a black box and it's not really useful to developers looking to change/improve the gpu performance.