Speaking about hackers, retention is probably underrated.
I have personally observed that a tremendous amount of software knowledge and understanding simply walks out the door when a really great designer/hacker leaves a team. The ability to simply ask the hacker who built some part of the function a quick question about the "big picture" rather than reading through lines (sometimes many, many lines) of code is a costly loss.
My extremely limited experience with acquisitions has been that employees at the acquired company are given large time-based retention bonuses. This is a market recognition of employee value. In some cases, this is kind of a waste - not everybody at a company is equally valuable. But there are always a couple of hackers on board who built the original codebase and you need to hang onto them if possible.
Of course, I've also seen companies acquired in what appears to be a strategic move to simply acquire the customer base too.
I believe the idea of employee value is at the heart of what Y-combinator is trying to do - recognize that some people can produce software that is capable of creating greater value than having that same hacker sit in a cubicle writing the next drop-down menu item in MS Word/Quicken/Oracle.
Further, there is always the domain-specific knowledge that walks out the door. We can read source code and eventually understand it, but the reasons behind the algorithms might be the result of years of experience in the domain. That's not easy to replace quickly.
The core team at my employer has been together for two and half years, and four members of the seven person team have been together since day one.
I feel like this is taking Zuckerberg's quotes out of context. I watched that interview, and more got the impression that he was simply pointing out the reality of the situation. Of course Facebook wants to keep great employees, and of course they try their best to do so.
The problem is that they know that great employees in a startup are, by definition, the people who always want to be working on the next big thing, and nobody is ever the next big thing for long. Facebook knows they took a lot of talent from Google when the next big thing moniker shifted their way, and they know the next big thing after Facebook (maybe Twitter, maybe someone not yet in business) is going to do the same to them. There's simply no avoiding it. You can only IPO once.
I didn't take his quote at all to mean they don't want to retain good employees or that you shouldn't bother. He's just pointing out that to a certain extent that's impossible and it's best to just admit that and work around it.
Also if I'm not mistaken Zappos is based in Las Vegas, and keeping programmers around there is a lot easier than SV.
Zappos also has fewer engineers relative to total staff. Lots of call center people, warehouse, logistics, where the work performance, management, and career paths are more straightforward than for programmers.
I think the postulation that most programmers like to hop jobs is incorrect. It's always been quite a hassle for me. Programmers, like others, prefer steady income. They prefer steady insurance and easy taxes, and all of the other benefits of staying at a job.
Facebook, afaict, is losing talent because its management is out-of-touch, overbearing, and ineffective. Google is losing talent because it's well on its way to becoming just another corporate entity.
People want to work somewhere fun and small. This usually describes startups. Google is now huge and is suffering some symptoms of that hugeness, and it's driving developers who were attracted to Google's startup culture out.
I think that it has much more to do with an acceptable atmosphere than "the next big thing".
Given two companies, one with above average turnover and one with below average turnover, What can we say? I think we can say very little without more investigation. The high turnover place may be terrible. On the other hand, it may be unafraid to hire really great people. It may try really really hard to keep them. It may be a tremendous victory for that company when a great person stays.
The other company may be very cautious about hiring. It may pass over great people that might leave, choosing instead to hire average to below-average people that have fewer options and little ambition.
There are other explanations, of course. I'm just pointing out that Mark and Tony might both be right, or neither might be right. It's all in the day-to-day execution of hiring people and managing them and running the business. Turnover alone is neither good nor bad in and of itself.
They also said those were their goals. They didn't mention their actual turnover. So while Zappos may have lower turnover than Facebook, it's because they're optimizing for that condition. Zappos is explicitly creating a culture of longevity.
I agree that Zappos is optimizing for low turnover. I cannot comment on Zappos specifically because I don't know them, I'm just pointing out that some companies optimize for low turnover in a very poor way, so I suggested that how a company achieves its turnover is more important than the turnover itself.
I suspect we are in violent agreement on this, just saying it in different ways.
As a current facebook employee, who is in no way empowered to officially speak for the company, I would not be in too big a hurry to extrapolate Mark's observations out to infinity. Facebook is putting together one of the very best software teams in the industry, and trust me, we care about keeping the band together. From my perspective in the trenches, facebook is doing a better-than-fine job of it.
I had a great conversation with another small business owner the other day. She runs a small design firm that he wants to stay small and nimble. She encourages turnover because more senior employees demand higher salaries and better benefits, and upsets the overall balance of power for such a small firm. So 3-4 years is the longest they let anyone stay, unless they're on a partnership track (which one staffer currently is).
Obviously the dynamic with is different for hackers, and especially those working on a single product over time, but it was a refreshing perspective.
My guess is that the answer is an unsurprising "It depends".
A high turnover can (emph. can) facilitate an upbeat, fast culture. Someone comes in, gives his fresh perspective on that particular problem-set and moves on. I sincerely doubt that Facebook doesn't full-heartedly encourages retention for key employees to ensure stable operation, testing and so on.
What you've described is called "consulting". It can be useful occasionally, but not very often. You don't hire on full-time employees as consultants without specifying that, it's a recipe for bad things. Employees expect a steady and fulfilling career; consultants consult, employees are employed.
Facebook is so backwards. Everything I hear about Facebook as a corporate entity bodes very poorly.
Employee retention is one of the number one most important things in any company, especially software companies, and double-especially in software startups.
A lot of executives and executive-types have absolutely no comprehension of the real cost of training a new programmer. If you have any kind of complex codebase, it's going to take _several months_ before a new hire can acquire ideal productivity -- and no, pair programming doesn't make this possible in a day and a half.
Startups have more than enough problems as it is getting something off the ground even in cases where all the founders stay with it all the way from concept to launch to profitability. Many startups whose original teams are intact die out before they become profitable.
Changing the lineup is just going to make things that much less likely, especially if your tech guy drops, because the tech guy knows all of the ins and outs of your actual product, which is probably undocumented and messy, and it will take a good programmer a long time to acquire the same level of familiarity with the product.
There are times when you don't have any other options to cut someone loose, but I would avoid it wherever feasible. Your company's essence is its employees; they run the company, their philosophies, management, and implementations are delivered to your customers, and they understand your domain, your products, and your accounts -- this isn't just intuitive knowledge that any qualified person (in whatever profession; accounting, sales, etc.) can waltz in and pick up, and retaining that wealth of information is critical to any successful business.
At least at most of the places I've worked, poor retention occurs when there is poor communication, disregard for an employee's cares and well-being, or a combination of the both. I don't understand why so many managers and executives have such a complex and insist on abusing subordinates, but it should be seen that doing so gets you nothing but angry ex-employees and ever-diminishing market share, relevance, and cashflow.
Fast turnover is such especial death for programming departments, and fast turnover almost always indicates serious systemic issues in the company.
Zuckerberg is crazy to think that this is legitimate strategy; it must be that he has seen the retention rates at Facebook and is now saving face by claiming it's part of some overarching corporate strategy so that the unprofitable Facebook can keep getting funding to maintain and purchase its servers.
This is not to mention that kick ass programmers that leave can end up being competitors. Any employee that's valuable knows the ins and outs of your company/software (what's good about it, where the limitations are, what customers want/don't want). letting them walk away is a risk.
Setting up an environment where employees often walk is even worse, i.e. where there is rampant "poor communication, disregard for an employee's cares and well-being, or a combination of the both". I've been there and watched great programmers leave unhappily. It causes morale issues and leaves a huge hole in institutional knowledge.
I have personally observed that a tremendous amount of software knowledge and understanding simply walks out the door when a really great designer/hacker leaves a team. The ability to simply ask the hacker who built some part of the function a quick question about the "big picture" rather than reading through lines (sometimes many, many lines) of code is a costly loss.
My extremely limited experience with acquisitions has been that employees at the acquired company are given large time-based retention bonuses. This is a market recognition of employee value. In some cases, this is kind of a waste - not everybody at a company is equally valuable. But there are always a couple of hackers on board who built the original codebase and you need to hang onto them if possible.
Of course, I've also seen companies acquired in what appears to be a strategic move to simply acquire the customer base too.
I believe the idea of employee value is at the heart of what Y-combinator is trying to do - recognize that some people can produce software that is capable of creating greater value than having that same hacker sit in a cubicle writing the next drop-down menu item in MS Word/Quicken/Oracle.