Google's definitely lost their way. Got too big, too fast and didn't preserve the culture.
The biggest an organization can get and hope to have a reasonable chance of not going evil is probably some multiple of Dunbar's number. I think realistically the uppermost limit on it is going to be Dunbar's squared and assuming everyone has really high social skills and no life outside work then it's 250*250 = 62k people. But that only counts if everyone's job is doing nothing more than keeping up with the company's culture; no actual work gets done.
If you assume that you want your people to work while they're at the office then Dunbar's probably goes down quite a bit. Let's use the more reasonable 150 number that's closer to average and let's assume that you only want people to spend about 20% of their time on keeping in tune with everyone around them. That puts your group size at 30 instead of 150 or 250.
Now let's suppose that we organize hierarchically and that the senior execs really do live/eat/sleep/breathe the company. They're professional, high functioning managers, so let's give them the full 250. And then let's suppose that there are group managers who split their time between talking with their group and with senior management.
That gives you 250 groups of 30 people each or 7500 total employees. And in order to keep up with all these groups you've got to have 3-15 execs who are all on the same page and running themselves ragged trying to keep up. If you start adding more levels of management hierarchy you run the risk that as information gets passed down the chain it gets modified slightly or tremendously to serve the interests of any one individual in said chain. And there are too many bosses you have to "go over" to get a reasonable chance of talking to a real decision maker if you think something isn't right, but your direct boss is telling you to do anyhow.
I'd guess that they did between 2002 and 2004. The Google+/youtube integration debacle started in what, 2013? That means it took them a good decade before they really started to do outwardly very unattractive things. That's pretty impressive (or my theory is totally wrong).
Either way, past a certain point more employees are more of a liability than a badge of honor. It just increases the risk you do something stupid because everyone believes they're doing the right thing and there are no lines of communication to correct the misunderstanding, intentional or not.
Everything I've read suggests that the G+ / GMail / YouTube integration was forced onto the various parts of Google by executive fiat: The rank & file Google employees were against it, but a company is not a democracy so it was forced on them & by extension us.
Executive culture tends to laud the "visionary leader" who pushes their vision on everyone else. When that vision is wrong-headed you end up with situations like the G+ débâcle: it was a failure of the Google executive, not the employees.
My guess is that we have the same effect in play with the YouTube Music debacle: an "up and coming" exec who wants to make their mark by creating a new business line for Google that brings in big profits. They don't care if they sully the Google brand in the process because their incentives aren't aligned with Google's, any negative effects on Google's userbase (be they musicians, music consumers or just ordinary users of Google services) are just collateral damage in their internal political game.
There would have been many ways that integration could have occurred, so that executive direction isn't inherently the problem - it's more if there was no one to direct the nuances of doing then there's a problem.
That kind of nuanced understanding is the difference usually between successfully companies and their unsuccessful companies.
I always feel there should be a single controller that reviews all major outreach tactics - or even a team dedicated merely to that once at that scale. It's just too important to skulk at.
Interesting analysis btw - thanks for sharing. I wonder if this is per company or per leader for trickle down effect - I imagine it is per leader, the one setting pace, vision, culture, etc.
My take is that separating the two is basically intractable and I wouldn't even try to quantify it.
One person doesn't have to work very hard to communicate all his/her ideas and priorities to him/her self so there's zero communication overhead there. A few executives are going to have to communicate a lot in order to all stay on the "same page" regarding the multitude of issues that a company might face trying to execute the overarching vision irrespective of it that's from one or a couple of people.
The more time these people spend communicating with one another to ensure that they all understand one another, the less time they have to spend with their group managers, thus lowing the number of groups they can oversee. As the number of groups an exec can oversee drops, you need more execs.
But as you get more execs, now you have a coordination problem at the executive level. The problem is recursive. The CEO can only oversee so many executives, so you get a CEO/COO/CFO triple team. But those people have to spend time ensuring that they all understand one another. Rinse and repeat.
Obviously there's a "management ratio" that a person could glean out of all of this and just employ that to structure an organization indefinitely, at least in theory. The problem is that all communication is lossy. If you can communicate with 85% effectiveness (which is PHENOMENAL!) then that kind of limits how many layers you can use.
0.85 ^ 2 = 72%
0.85 ^ 3 = 64%
0.85 ^ 4 = 52%
0.85 ^ 5 = 44%
0.85 ^ 6 = 37%
This sort-of tracks with real world businesses too, it's not often that you see an org chart which has 10 layers on it because the company would be totally dysfunctional and go bust very rapidly.
The "Peter Principle" applies to businesses too, the continue to grow and gobble up more market share until eventually they stop being as effective and growth slows or stops.
It applies to society as well as Joseph Tainter points out in "The Collapse of Complex Societies" each additional unit of complexity has a fixed cost and diminishing returns for the society (or the company, or whatever) until eventually you're adding complexity which is a net negative. Things then start to go downhill.
Obviously there are exceptions to this (think "natural monopolies" and the like) and there may be years or decades of lag between the start of net negative complexity growth and collapse. Unless there's "refactoring" of the system where complexity is removed somehow, everything will eventually trend towards collapse. And achieving refactoring in real life is generally impossible since there tend to be entrenched interests in favor of some complexity since it directly benefits them.
A great example of this is the healthcare debate. Doesn't matter which side you're on. Complexity nearly always wins.
I think communication structures should inherently be clear enough to pass all critical information over, and if there become communication issues or misunderstandings then you have to eliminate that communication break - whatever that means.
I feel some of these issues can be managed by having specialized teams - but those teams would have to managing a context, not merely certain functions or tasks of a business, otherwise they won't have their hands and attention on everything that will matter.
85% communication effectiveness seems very possible - at least with enough time and enough opportunities.
The biggest an organization can get and hope to have a reasonable chance of not going evil is probably some multiple of Dunbar's number. I think realistically the uppermost limit on it is going to be Dunbar's squared and assuming everyone has really high social skills and no life outside work then it's 250*250 = 62k people. But that only counts if everyone's job is doing nothing more than keeping up with the company's culture; no actual work gets done.
If you assume that you want your people to work while they're at the office then Dunbar's probably goes down quite a bit. Let's use the more reasonable 150 number that's closer to average and let's assume that you only want people to spend about 20% of their time on keeping in tune with everyone around them. That puts your group size at 30 instead of 150 or 250.
Now let's suppose that we organize hierarchically and that the senior execs really do live/eat/sleep/breathe the company. They're professional, high functioning managers, so let's give them the full 250. And then let's suppose that there are group managers who split their time between talking with their group and with senior management.
That gives you 250 groups of 30 people each or 7500 total employees. And in order to keep up with all these groups you've got to have 3-15 execs who are all on the same page and running themselves ragged trying to keep up. If you start adding more levels of management hierarchy you run the risk that as information gets passed down the chain it gets modified slightly or tremendously to serve the interests of any one individual in said chain. And there are too many bosses you have to "go over" to get a reasonable chance of talking to a real decision maker if you think something isn't right, but your direct boss is telling you to do anyhow.
This chart shows that Google cracked 7500 employees before 2007. http://www.statista.com/statistics/273744/number-of-full-tim...
I'd guess that they did between 2002 and 2004. The Google+/youtube integration debacle started in what, 2013? That means it took them a good decade before they really started to do outwardly very unattractive things. That's pretty impressive (or my theory is totally wrong).
Either way, past a certain point more employees are more of a liability than a badge of honor. It just increases the risk you do something stupid because everyone believes they're doing the right thing and there are no lines of communication to correct the misunderstanding, intentional or not.