I can't be positive what the parent post was referring to, but it's likely it has something to do with re-balancing the dev/test/pm/manager roles. Microsoft had started to get top heavy, so Sinofski had a lot of managers demoted to individual contributors and invented this odd "triad" system of dev/test/pm to try to keep things more balanced. It struck me as a passive aggressive and politically correct way to force some under-performers out. It had the positive side effect of preventing runaway PM orgs, but had the negative side effect of encumbering some well balanced teams. I assumed that it was an interrum strategy to get a handle on organizational complexity. However, other orgs around the company started being "Sinofski-ized" without any real understanding of what that meant.
He was a big fan of the triad- Dev, Test & PM, at every level. From what I experienced, at every level you'd have to get sign-off from all 3 for any features to be implemented. That, combined with Microsoft's incredibly deep org structure created a massive number of 'committees' to go through to get signoff for any work to be done.
He was also one to dictate things from up above and it was extremely difficult to understand the reasoning behind them or offer any form of disagreement. Anyone who didn't follow exactly what he wanted, was out.
So basically you'd have committees of 3's (Dev, Test, PM) filtering and relaying every decision. The people who could work the politics would get promoted and the people who understood the details would get frustrated by the top-down ambiguity and falter or leave.
A friend of mine (another ex-softie) described the genius of modern day Microsoft to be taking C players and reliably producing a B product. The triad model seemed to fit right in line with that ideal.
IMO the sign-off culture would be a problem at MS with or without triads. When you have such a large middle-layer (especially ones with ranks like 'partner' who seem to mostly be faking it) there are going to be loads of people whose chief role is to be a gatekeeper who needs to 'sign-off' on something. There are going to be layers of management who want to perform that filtering and relaying you mention, even though it does nothing for the company. And until some kind of purging happens, that won't change.
Honestly it wasn't so bad. PM's with vision and proof could force their will, devs with talent and credibility could dictate the approach, and testers with open eyes could hold the team to their commitments. It's ok to specialize and lean on your partners.
Microsoft was all about this while I was there. It was incredible how much permission you needed to do anything. When I left I found an almost uniform response from other ex-MSFT employees that they "just couldn't get shit done" while there.
This is probably how things work in most big old software corporations. It is not open to big surprises, like a revolutionary new product, or a catastrophic failure. If the company is on a profitable turf, I would say, it is the way to go.
For individuals though, it makes the utterly limiting environment, where gatekeepers and not the ones with merit flourish.
On the bright side, my unscientific observation is that, in the best case, such an organization can go on like this only for one career time (~20 years). So, if you are coming in towards the end of that period, you are in for an adventurous ride.