Was Valve ever "structureless"? My impression from reading the handbook was that power structures were recognised but not enforced, i.e. if you wanted someone on your project, you had to persuade them. This is not great if you can't get people to work on your pet project, but that doesn't mean the system isn't working.
Personally, having seen Jeri's AR project video, I wouldn't have joined it. It seemed like a rather weak and non-transferable technology that didn't solve fundamental problems with AR and that had limited gameplay possibilities.
What Jerri has reported is a step past that, where it wasn't just about attracting other interested parties, but a combination of being denied resources to hire folks who had the needed skills but didn't fit the valve stereotype, as well as a small group within valve actively and aggressively seeking to purge people they objected to.
If you look at some of the other famous horizontal management companies like SEMCO, they have a periodic bidding/approval cycle. People make a proposal, ask for a budget, and once granted it have temporary autonomy. The counter-pressure is that however they spend their budget is tracked and transparent to everyone. Hence if you fuck around and blow your whole budget on perks for yourself, your proposal won't be continued next time it comes up for bid. From what I remember SEMCO found 6 months a good period.
This is obviously speculation from limited information, but taking Jeri's comments at face value (which I think is entirely reasonable based on how she's conducted herself publicly in the past) valve seems to be missing that key element of temporary autonomy.
An environment where the socially powerful can raid or eject you at any moment they wish is just going to amplify internal political dysfunction.
It seems to me that the organization worked as it intended to work. The consensus was the project wasn't a big enough moneymaker, so they didn't fund it properly.
Of course if there was a boss, there would be a pitch, a cold, hard look at the budget and then a decision, which you may or may not like, but you would have to live with it. At Valve there's no one person to blame that you didn't get what you wanted.
But the result is the same: The company didn't go forward with an interesting but not-obviously profitable project.
She was short of people inside the company who wanted to do the job. Of course you can always find people outside the company who want in (it's a job, after all). In the case of hiring people specifically for someone's project, presumably the hiring process has to act as a proxy for the 'is that project worth working on?' or inflexible resources (a machinist isn't going to work on Steam) get tied up in a project no-one else is interested in. Hiring the person as a temporary non-employee contractor would probably solve that, but I see how Valve's resource allocation system might make that difficult.
And they already had purchased over 100K of equipment, and yet nobody could use it since they couldn't hire someone who knew how to use the machinery - like a machinist.
Personally, having seen Jeri's AR project video, I wouldn't have joined it. It seemed like a rather weak and non-transferable technology that didn't solve fundamental problems with AR and that had limited gameplay possibilities.