[sebastianbk]

As a Microsoft employee I am so happy that we got rid of stack ranking a few years ago. It encourages a bad behavior and goes against helping your coworkers with whom you are essentially competing for compensation. I am surprised to see that a company like Valve, which seems to be held in high regard by many developers in the industry, still operates with this compensation system. It's system of the 80's if you ask me.

[AimHere]

I suspect it might work differently in Valve's case. I'm led to believe that Microsoft had a fixed, conventional hierarchy, where every little group and person within the group was backstabbing everybody else to keep their jobs.

If their handbook is to be believed, Valve has a much more flat management structure, where it's basically Gabe at the top, sortof, and everyone else doing whatever they think is best for the company, and there's a fluid system where people can move between groups according to their interests and how they perceive they can add value. So, unlike in Microsoft's case, Valve's people have an easy avenue towards putting 'if you can't beat 'em, join 'em' into practice.

Valve has a radically different corporate culture from most other companies in it's space. It doesn't come from the 80s, or indeed almost any other time. Perhaps the stack ranking works a lot better because of it.

[Karunamon]

Part of me wonders if the Valve management structure isn't almost completely to blame for things like their infamously bad customer service.

I could see why - dealing with support tickets from irate people is not a particularly interesting (or judging by Steam's runaway success: particularly value adding) activity.

Not that I mean to hijack this to complain about Steam, but you have to admit it's a benefit of a traditional management structure: someone is making sure the shitty-but-necessary work gets done.

[AimHere]

I suspect it's much more that Valve is an insanely lightweight company for it's customer base. It has something like 330 employees servicing 100 million customers on the go-to PC gaming platform, and a lot of those employees are working on shiny tech and new games. As a comparison, Rockstar North's facility in Edinburgh has roughly the same employee count (slightly more by Wikipedia's count), and all they do is put out one new game every 5 years or so.

I suspect that if Valve wanted to, it could easily hire lots of people who would be happy to man the customer service helplines and who wouldn't be either able or willing to try to make TF2 levels or whatever.

Maybe the management structure is what limits the size of Valve, though.

[Keyframe]

Rockstar North's facility in Edinburgh has roughly the same employee count (slightly more by Wikipedia's count), and all they do is put out one new game every 5 years or so.

All they did was recreate a city full of assets and added game mechanics to it in five years and made the most expensive (at the time) game product in history with more than 1000 people involved in production and the fastest selling entertainment product in history. A game that broke eight Guinness world records and sold more than 65 million copies and well over $2b in revenue. And that's only GTA 5.

All they did, numerous times so far, was to create vast games full of created digital assets. Games that hold record-breaking sales numbers and hold positions in all best-selling video games charts. That's all they did. Nothing to write home about, really. Bunch of slackers.

[AimHere]

What's your point? Valve put out multiple AAA games in the same timeframe too, and did a lot of other stuff besides. Sure, GTA5 is a big-ass game, but Valve's games - games plural - haven't exactly been small projects either.

[Keyframe]

DCC of assets for something like GTA is magnitudes larger than anything Valve does.

Rockstar has an army of artists working on those assets for that game. Valve is a multidisciplinary house where they dabble in a lot of stuff, split in smaller groups, but their primary vehicle is Steam and a couple of (successful) games where each project does not need as many people. It's comparing apples to oranges.

What Rockstar does demands an army of artists. If Valve were to do same type of game world, they would have an army also. They do not do the same thing though.

If Rockstar were to do what Valve does, they would also have a number of people split into smaller teams working on their own things. They do not do the same thing though.

That's why you can't say it's all Rockstar does with same number of people. Those people are not the same skill and those projects are not the same.

Rockstar's yield, if you're comparing the game industry, is a level above Valve. It's a level above most. It's really not a good example to compare to. They are on the leading edge of what they do. An area Valve isn't in. Maybe if you've compared Avalanche Studios and Just Cause 3. They had 75 core-member team in a studio of 250 and, who knows how many, outsourced people.

[Kurtz79]

His point is that probably just GTA5 has sold more copies in 5 years than all Valve games combined (a quick check of the most popular ones, HL 1/2, L4D 1/2, Portal 1/2 reaches about 30 mils), so that number of employees is justified.

Also:

http://www.gamespot.com/articles/gta-5s-online-mode-has-gene...

There is no denying Valve does great with the number of people it has, but of all the badly managed companies they could be compared to, Rockstar really it is not a good example...

[SXX]

Pretty sure Valve outsure all customer support.

> It has something like 330 employees servicing 100 million customers on the go-to PC gaming platform

Actually there is a lot less people who actually have any connection to userbase at all. Most probably never even publicly known.

Few years ago one producer who visited Valve commented that in office there was one of areas where he was told something like "and here is our Steam department". And back then that was like 20-25 people at most include both partner relations and marketing.

[frik]

> shiny tech and new games

You mean Half Life with it's last full iteration from 10 years ago, and the second episode with a cliffhanger from 2009? Or the Source engine (HL2) which evolved from GoldSource (HL1) which was based on Quake 1 engine (with some parts from Quake 2 engine) - even the Call of Duty engine which aged well too is based on the more modern Quake 3 engine, and I wouldn't consider it shiny in 2016. Rockstar's RAGE engine as well as Crytek engine (StarCitizen, Crysis), and the Frostbite (Battlefield 1) are light years ahead and far superior in every aspect and that's what I would consider shiny tech.

Valve is what Valve is today because of the success of steam. They produce little games like Dota2 more for fun than anything else, they gt rich with steam.

[AimHere]

It's a bit disingenuous to just mention Half Life, although note that work on Half Life series games was still ongoing as of last year (and they did a fair amount of work on the whole series in ~2012/2013 porting it to other operating systems).

Those 'blank' 10 years saw the release of Left 4 Dead 1/2, Team Fortress 2, Portal 1/2 and CSGO, for instance, as well as that 'little' Dota2 (which is clearly a large game by any metric you care to choose). At least three of those games involve regular ongoing content infusions. Add in the development of the Source 2 engine, and the Steam Controller, and the work on the SteamOS platform and whatever else on Steam Machines.

This is out of the 330 employees that also support those >100 million customers using the service on a regular, possibly daily, basis. As I pointed out, Rockstar uses more manpower than Valve just to create essentially one game every few years. The last GTA game I bought (San Andreas) did have graphics well behind the state of the art too.

Even if we assume that those 330 employees are doing nothing but working on Steam, it's still a tiny number compared to the customer base. AirBnB, for instance, has had 2 million listings in it's lifetime, and it has about 2300 employees. Uber has over 6000 employees on about 8 million customers, and both those cases are ones where people aren't typically regular users.

[huac]

the majority of those airbnb/uber numbers are customer support not engineering

[izacus]

None of the "blank" games are Valve's IPs but acquisitions which kind of stopped in development or releases after Valve acquisition. Half-Life was their last original IP and even that ended in a complete cliff hanger with Episode 3 or sequel nowhere to be found.

It seems that Valve has serious problems actually pushing their games to completion since they started printing money with Steam and pressure to actually release went away.

[antisthenes]

> They produce little games like Dota2 more for fun than anything else, they gt rich with steam.

Little games like Dota2? It's the biggest title on Steam in terms of players by a large margin and probably the 2nd most popular game worldwide on PC to ever exist (behind LoL)

It also generates a significant amount of revenue for Valve through the sales of tournament ticket and cosmetics.

[squeaky-clean]

You don't mention Portal 2, Dota 2, CS:GO, all the TF2 updates, Source 2, all the advancements to Steam, the Steam Controller, Steam Link, the Vive, the Lab VR demos.

Sure the Battlefield team has made Frostbite which is shinier than Source. Did they also make their own store, gamepad, and VR headset?

[theinternetman]

> the Lab VR demos

Wii minigames in VR are still just Wii minigames, whole thing feels like something I'd expect from a hackday from a company as supposedly prolific as valve.

Got a strong feeling the talent is long gone.

[Johan-bjareholt]

Source 2 is in DOTA 2, but that game doesn't show its full potential. Let's just wait for their next game until we say that they are "years behind".

[danso]

[insert obligatory Half-Life 3 joke/reference here]

I've never used Steam's customer support; do they outsource any of it to contractors and/or offshore?

[wishinghand]

They've tried and they discontinued doing so because they felt it was even worse than being apathetic about it.

[Sebguer]

What's your source on that? I'm fairly certain that all of Valve's customer support is outsourced, though not specifically overseas, but to contract support teams.

[theinternetman]

> Rockstar North's facility in Edinburgh has roughly the same employee count (slightly more by Wikipedia's count), and all they do is put out one new game every 5 years or so.

Yeah they actually manage to ship games (that manage to make back the entire production budget in pre-orders alone), not just sit on their hands running an online store with an outdated and clunky client.

[tikhonj]

I mean, it's not like traditional companies—from Google to Comcast—consistently provide great customer support either. So you can't really make strong conclusions about why Valve is having issues there without additional insight; there are too many confounding variables to pin it just on its management structure.

This is a common story with people trying something new. If you do the old-fashioned thing—buy IBM, so to speak—and fail, well, these things happen. A lot of factors could have contributed to the problem. But if you try something new, the new thing must be at fault—even though all those other factors apply just as much now as they did in the IBM case.

[Karunamon]

That's a fair point, but I'm struggling to think of a way in which it isn't at least partially to blame. The handbook (and Valve insiders) say it's pretty much a do-your-own-thing company. Who'd want to deal with angry users all day?

[hodwik]

I work for a software company very much like Valve in structure, who has been operating this way since the 70s.

Support still gets done, because when you hire you specifically hire people who love doing support. Those people exist, and they get tremendously emotionally involved in the quality of their work, just like anyone else.

In every department there are people who struggle with the flat hierarchy and free range to work on what you like, who have a strong emotional need to know who is in charge, and to be told what needs doing. Those people struggle, but they are by no means relegated to any one department -- plenty of them are engineers.

[robbiemitchell]

If they increase the weekly/monthly pay for people who choose to deal with angry customers, it should reach equilibrium eventually.

[Bjartr]

Surely there are some people out there who really do have a passion for, and get satisfaction from, dealing with upset customers. The questions left are how hard are they to find and how expensive are they to employ.

[touristtam]

Or someone is making sure to look like their are doing some work. I have seen plenty of managers, that are nothing more than a human layer between the CEO/CTO and the engineers.

[iEchoic]

I'm a former MSFT employee that was around when we dropped stack ranking, and I didn't feel like that was a substantial change. Managers still calibrate you against your peers, a stack is still created, and compensation is assigned accordingly. I remember reviews feeling the same before and after. What changed for you?

[localhost]

At any company (Microsoft included) where performance-based compensation exists, there is a budget for that line item. Therefore it is a zero-sum game - to pay someone more because of their performance, that means that someone else gets less (or zero) from that line item.

So, depending on how you interpret the term "stack ranking", you can either look at it as "forced removal of the bottom x% of the company" or "people aren't ranked / bucketized". In the MSFT case, I believe that the former has been removed. But the latter definitely cannot be removed if you are to have performance-based compensation.

[runeks]

    > At any company (Microsoft included) where performance-based 
    > compensation exists, there is a budget for that line item.

I'd like to point out that it's possible to pay bonuses out of profits, rather than a fixed, yearly pre-allocated pool. If employees are only paid a bonus when they add to company profits, there is no competition for bonus funds between employees. So for example, if you have a great idea that saves the company 30% on something, you get half of that saving/profit, or something similar.

[cesarbs]

The problem is, how do you measure that? There are a bunch of good ideas that most people will agree provide a benefit, but it's really hard to come up with an objective measure of how much money that makes or saves.

[crpatino]

Make the bonus pool a fixed percentage of company yearly profits (not sales). Assign at least half of the bonus pool in an egalitarian way (like based on the number of days worked on that year), assign the remainder on a roughtly "merit" based criteria, and let every team to agree on the criteria, but make it be something objective (it does not matter if it is number of bugs closed, or being at your desk on time in the morning, or whatever, just as long as the measurement is unambiguous and the team agrees on it).

That would be far from perfect, there will be free loaders, and most likely than not it will be slightly unfair to everyone. But at least you have removed the perverse incentive to sabotage coworkers in order to look yourself better.

[runeks]

If there is no measurable difference, then a bonus cannot be paid out according to this scheme. So we're talking about innovative ideas that release an extra reward (in addition to a regular salary), if they result in a measurable saving somewhere.

For example: you write an algorithm in your spare time that more efficiently packs together a good that your company produces, such that shipments take up 10% less volume thus saving ~10% on shipping.

[antnisp]

You will be horrified to learn that there is a push to write stack ranking for all public employees into the constitution of Greece.

[dpweb]

Wouldn't it make sense to just pay everyone (in the same type of role) essentially the same amount and then bonus people from time to time on particular outstanding achievements. Nothing is a more powerful inducement than financial incentives and a bonus gives an impact that the ongoing salary doesn't. People don't give a crap about annual reviews unless they think its low enough to get them fired.

My sense is performance evaluations should be banished from the corporate world, for the most part. Usually a waste of time, but that is where managers can be helpful as they are carrying an ongoing assessment of the value of each of their employees at all times.

[ck425]

'Nothing is a more powerful inducement than financial incentives'

Actually there's a lot of research that contradicts this. Essentially as long as people have enough money that they don't worry about money ie a comfortable middle class lifestyle for country/area, and they don't think they earn significantly less that their peers, money is very inefficient and occasionally negative incentive for tasks that require a decent level cognitive ability.

[thenomad]

IIRC Peopleware showed that financial bonuses were tremendously effective - at reducing peoples' investment in their work.

[usrusr]

Extra money for excellence comes with an implied message that the regular money is just for showing up. You really don't want your employees to feel that way.

[f2f]

that guide is from 2012. things may have changed by now. i know we gave up stack ranking at $GIANTCORP since then.