Nick Cook, former creative director at Pumpkin Studios, looks back on the industry he left…
"During a 21-year career in the games industry, I’ve occupied every position from production artist, to the senior management of Pivotal Games and Pumpkin Studios. I have also worked on the publishing side during my time at Microprose.
I’ve seen the crunch problem from every angle, but it wasn’t until I left the industry four years ago that the benefit of hindsight really began.
To put my cards on the table, I believe that extended crunch time is simply unsustainable.
Not only does productivity sharply start to drop, but also quality begins to suffer … and that’s just the consequences at work. I feel far more serious and corrosive, is the human cost of all this. An extended crunch saps morale, destroys work/life balance, burns people out and affects health.
Its continued existence shows that something is fundamentally wrong in the production models that most studios use.
Advertisement
I know how easy it is to point the finger in this situation, but as a HR manager once advised me, blame culture serves no one and ultimately, changes little.
Everyone gets defensive, hackles are raised, and scapegoats are selected. Yes, without doubt certain individuals are probably more culpable than others, but everybody has their part to play in contributing to a situation where crunch becomes inevitable.
Many production staff find it practically impossible to predict how long a task will take; producers are often too quick to believe an unrealistic estimate and not allow sufficient slippage in the schedule; studio managers keep changing their minds and want ever greater bells and whistles squeezed in; then of course the publisher often contradicts studio management direction and still insist that completion dates are hit regardless. It’s a recipe for disaster and its ultimate consequence is crunch time.
So is this pit of doom really inevitable, or can something be done about it? Well, here are my thoughts. You may not agree, but if nothing else this will stimulate your own thoughts on the matter and encourage debate that leads to real change, because that’s exactly what the industry needs to do. Extended crunch time was wrong ten years ago and its still wrong now.
A VERTICAL SLICE OF PIE IS NOT ENOUGH
It makes so much sense on paper doesn’t it? An example level that shows off all the major features and USPs of the product. Publishers get enthused… they can really shift mega-numbers of this new baby… sales estimates drawn and financial models constructed.
The studio gets the green light for full production and the cash starts really flowing, everybody is happy – until problems begin to mount up. It all starts falling apart: the inevitable feature creep has kicked in; wobbly tool sets keep breaking; bug counts go through the roof; and art quality that looked great for the demo, suddenly is way behind the curve of the latest products hitting the shelves. This tale of production woe has a familiar ring to it and is the industry’s very own, Groundhog Day.
Don’t get me wrong. I believe the vertical slice philosophy almost works but fails in one major area – it’s not ambitious enough. Forget a slice of pie, we want it all. A prototype should be the whole game, something that uses only rudimentary graphics to keep costs down, but is fully playable.
If there is anything that should be proved before setting off on full production, it’s gameplay. As an art director I was often dismayed by how often people would fixate on the eye-candy at this early stage. Yes I know it gets sales and marketing pulses racing when a slick graphic demo before them, but is it really going to get great reviews for gameplay?
Certainly you should produce a proof of concept for graphics, but don’t fixate on it. A prototype should be robust enough to lock key content down and to give sufficient insight into a product’s potential for the publisher.
This has to include any major coding and scripting requirements. It is this prototype, design complete (stress test with dummy graphics as necessary), that is signed off on.
The next stage is the key part… this is what is then built. Kill off feature creep by extending the prototyping period as long as it takes. Do not let full production start until you know its going to be a great game.
STOP MAKING IT UP AS YOU GO ALONG
I can hear the gnashing of teeth already… what about the costs I hear you say? Well, this is the beauty of extending production time for a prototype… do it with a skeleton team.
The size of that initial team will have a lot to do with the ambition of the project, but by keeping the staffing relatively small, costs are minimal compared to a full team.
And the real beauty is you keep the prototyping period going as long as necessary. Publishers, you have to view this as a loss-leader. Fund lots of prototypes in the hope that one of these frogs may turn out to be a princess. Why? Because you only throw the big money at something that’s really proved itself with rudimentary visuals. You know it’s a great game because it’s actually past the smell test.
But the deal has to be this is the game that’s going to be built. Think of the prototype as a car designer’s blueprint… it’s the finished design. When it hits the factory, people don’t suddenly decide that the SUV they’ve agreed to build, is actually going to be a sports car, and start adding bits as its slides along the conveyor belt.
IS THE STUDIO MODEL TO BLAME?
Costs are escalating as the complexity of games demands ever greater team sizes. By placing a large studio onto a production footing from almost day one, burn-rate is massive.
Does it really have to be this way? Look at Hollywood. Only once a script is green lit and the director has been selected, will the actors be cast and production resources drafted in… and they are freelance. When a film is finished, the team dissolves back into the talent pool.
Couldn’t we do the same? A small core prototyping group dovetails rather neatly with the idea of surrounding them with freelance resources. Hire in the staff once the project has got the go ahead. Like the film industry, it would be easy to imagine lots of independent specialist companies forming. A design group with a proven track record in interfaces, scripting specialists, that guy who did amazing lighting job on that other game – you get my drift. Yes, of course you will always need a core team, particular in areas of code and tool sets… but this at least worth thinking about and is directly linked to squeezing crunch out of the system.
The out-sourcing model is most effective when they are given a rock solid brief. That’s exactly what you have with your full prototype and will help them hit their deadlines. Suddenly a lot of the guesswork goes out of the system, i.e. scheduling, and everyone is dealing with a known quantity. Perhaps the real future for studios will be that they become the core prototyping team, surrounded and augmented by a much stronger freelance industry, populated by a lot of the talent currently working in-house. I certainly believe this would take a lot of stress off the studios, reduce the hiring and redundancy cycle that’s once again rearing its head. Most of all I believe it would reduce crunch time.
SUMMARY
I know there are lots of ifs and buts that you’ll be coming up with, but these are just my ideas and the intention is to stimulate debate. Crunch time can probably never be killed off completely, but many studios can certainly handle it more effectively than they are now. Just think about this for everyone’s sakes, including the profit bottom line.
You develop in a fast moving technologically based industry, so think radically in terms of your solutions for it. My fundamental philosophy is that great games come from happy staff. Create an industry where people can work and play, live lives that are in balance. The time to ask questions and come up with new sustainable strategies, has come. "
You cite some of the biggest problems with the current studio/publisher development model when you talk about improper planning (someone accepts pie-in-the-sky estimates), too-soon production with full teams, and publisher-induced changes. I think the small core team + large amounts of contractors is an interesting suggestion, but of course the biggest hurdle there is a serious retooling of the current labor arrangement. Lots of people would have their previously solid jobs pulled out from under them. Would they be able to find reliable work quickly enough as contractors? Who would be willing to foot the bill for the costs associated with the retooling period? It wouldn't be the publishers, I'm guessing.
I don't mean to be all doom and gloom. Kudos just for sparking discussion.
Speaking as a programmer, doing 'a vertical slice' is 'doing the whole game'.
The rest is just assets.
Yes, ruin it even more than it is ruined now.
I am in my late thirties, I have two children. Relocating frequently is not an option. I am probably representative of most senior developers. It's alright for twenty-somethings just out of university to up sticks and move to another part of the country every year or so, but for us older folks it doesn't work.
If the games industry starting contracting as standard, I would take my skills out of it. I work in the games industry because I love making games - but I can get paid more in the 'real world' and have greater security.
@CdrJameson: then you must be an inexperienced programmer. A "vertical slice" is never even close to the finished product from a code point of view. Why? Feature creep! Goes something like this... vertical slice is done, so some producer/designer/marketing/publisher guy says "the slice is "interesting", but its missing some core elements that would make it much more fun, let's go into production but add A, B and C". Result? Lots more coding!
Does the "prototyping with skeleton crew" approach work when most games are sequels or movie tie-ins and have to be out by either the movie date or the last day of a financial quarter? This is not like a movie script which can be perfected before production starts.
@Martin: the current contracting model we see now (with the contractor working in-house) is not necessarily what Mr Cook is advocating.
It seems to me that he is advocating that studios have a small core prototype team, rather than a massive burn team production team in-house, and then work with specialized companies, that have their own studios, to do lighting, assets, UI, sound, scripting, etc.
So, you would either be working full-time in a development studio or working full-time in a outsourcing studio. How is this upping sticks?
@Bob: that kind of model sounds feasible, and as a programmer I wouldn't be outsourced anyway.
However, my experience in engines/tools tells me that this method of outsourcing could be difficult to implement. Contractors will need access to build systems, runtime environments, asset repositories - and none of the companies I have worked for so far have been capable of doing this. If you ask contractors to work blind - i.e., just create Max models on their own - you are asking for trouble. I've seen this happen. Without a contextual environment, assets usually don't 'fit'. Then you have to get the contractors to do the work again. Additionally, you need people at base to verify these assets, integrate them into the in-house resource system, etc. That kind of work is not trivial.
@Bob
Maybe that's where schedule overrun comes from: Bored coders!
Seriously though folks, a computer is a machine on which it is expensive to do something the first time, but you get any number of copies/uses for nothing.
A vertical slice is... doing everything the first time. It's precisely the least efficient way of doing anything.
No wonder it ends up buggy.
A few of the comments here cite problems of moving over to a studio/contractor model, but these are all problems which already adversely affect the the development of games anyway. IMO it's a bad idea to reject a new model merely because "we don't do it that way". Shoring up a system that is not working (and let's face up to it, it's NOT working) merely because it would take work to get it going properly is not the best course.
Many people refer to the film industry and they managed to make the transition to outsourced contractors at about the same stage of the industry's development that the games business is at now.
Time to bite the bullet, I think.
After 20 years as an artist in the games industry I'd like to offer my observations for a major cause of crunch, this is not just gawping around the art room, I've tested this by tracking productivity and doing A/B testing across different teams at the same time and having repeated this test many times, a bit like Henry Ford did a gazillion years ago.
When an artist (and presumably coders and designers have similar situations) starts a new block of work such as a new level, there is a lead in time, a getting up to speed time. By comparison when finishing off a block of work the same artist is capable of true superstar production rates, that is until the work block is finished and a new level (or whatever) is started. So making assets one after another, each one different requires some retooling of the mind and also of the palette of assets to draw on and production speed is like a sine wave. Sounds obvious? Yes it is, but in my experience schedules are not written like this, schedules are written so every equivalent task is given the same amount of time and these time blocks are then thrown around and assigned to different artists willy-nilly, so obviously a situation arises where the schedules become a work of fiction.
The second part of this problem is how individuals react to this. It's not nice being behind schedule, having the company bulldog at your heels, having that your-bonus-is-getting-slimmer look as the producer adds days of time to your schedule because the widget you made took 3 times longer than the schedule stated. We feel obliged to complete and compete with the schedule (which others are not struggling with, see the previous paragraph for the reason) so inevitably an evening or two extra will bring you up to speed? Wrong. I've tracked teams and individuals who have done this and what happens is - they never catch up, as they get more and more tired they get slower, make more mistakes and eventually the workload gets shifted to somebody else to help out and so they also become overloaded, do some overtime and before you know it you're in a crunch spiral. Then the blame starts, the managers get stressed, the schedules slip further and moral plummets.
As I stated above I've tested this and teams who don't start doing overtime finish faster than those who do. They start to get behind at first, and then catch up about 3/4 of the way in, in addition they are able to cope with a little feature creep as they are fresh throughout the work block and are by this stage, super fast in the context of the job at hand. Compare this to a team doing overtime, they start to catch up initially but this drops away quickly and it takes them that much longer to get up to speed so each individual task takes longer and in the end they crunch and are still late.
Another more subtle effect is that if you've not crunched to finish an asset, you like the asset it has a positive emotional context so when you go back to it for changes or whatever you can get back up to speed with it that much faster. By contrast an asset which has been made during crunch is a hateful thing that you have to slowly drag yourself back to.
So although Nick brings up many good points on avoiding crunch, low level scheduling in a sensible way that expects people to be humans not robots (and yes, it will take more managing, and the producers have more up front work) will help enormously.
What bugs me the most is that this is common knowledge to anyone who's studied business (such as producers and studio heads) and yet many are happy to let people dig their own crunch graves by repeating this pattern year in, year out.
Martin. There is very little job security in the games industry, studios close due to missed deadlines and pulled projects, finish the game and sack the team is also becoming increasingly prevalent.
It's been drifting in that direction for years and started long before the current financial crisis.
It's going to start heading towards something resembling the film model anyway.
Some aspects of production have been doing this for ages, (music/audio for example)and for bigger titles, the cheapness of overseas labour makes it very attractive when you have thousands of props to get built.
