Scylla, Charybdis and Atropos

An article entitled [url=http://www.gamecareerguide.com/features/482/how_halo_3_changed_game_.php]"How Halo 3 Changed Game Development"[/url] has recently drawn some (rather sceptical) attention [url=http://halo.bungie.org/news.html?item=21601]at HBO[/url] and here [url=http://rampancy.net/story/bungie/10/01/2008/Did_Halo_3_Change_Game_Devel... Rampancy[/url]. The author's enthusiasm for the Halo series is unquestionable, but the article's relevance to its title is definitely open to doubt.

Rather than pandering to my inner sadist by shooting the original article to pieces, I'm going to try something more constructive. As someone with relevant experience, what do I think Bungie has to teach the rest of the industry? I can't promise any stunning revelations, but even the obvious bears repeating from time to time.

The single largest problem in games development today (in my not-so-humble opinion) is that of effectively managing and co-ordinating the efforts of the huge teams required. I believe Bungie's in-studio team for Halo 3 numbered about 100. I used to think that it was simply impossible to maintain flexibility and creativity in a team of more than 20-30. My hat is very definitely off to those who have proved me wrong, but keeping a games development team as small as possible is still a very desirable goal. It's interesting to note that Wideload, Bungie's sibling, seem to have taken this approach.

Project management for ambitious games software is one of the most difficult and delicate jobs imaginable. To do this properly, everyone involved has to steer a course between two undesirable extremes:

A) Inflexible bureaucratic rigour - a design-by-committee, built straight from the spec. sheet with no deviation allowed, and probably still not delivered on time, not least because that spec. sheet has to be written - and re-written - in ever more mind-numbing detail, and everyone spends half their working day filling in timesheets and sitting through progress meetings.

B) Creative anarchy - lots of bright ideas, constant fundamental revisions, and a completion date forever receding into the future.

B probably used to be the more dominant pattern in the games industry, but these days it's very definitely A. Readers should be able to identify examples of both approaches. Observant readers may have noted that the one thing they have in common is late delivery - or on-time delivery of a badly flawed game. I said it was difficult.

Yet work simply has to be organised, and progress monitored. The people putting up all that money have every right to be kept informed of progress, and these days a publisher practically has to buy the retail shelf-space for a new game (and well in advance, at that), so shipping dates must be met. Every really good AAA game published represents a successful navigation between these two deadly whirlpools, while racing for the finish line. Bungie aren't by any means the only company who can do this, but the magnitude of their success with the Halo series means that they attract more attention.

It's easy to describe the problem - so what's the solution? Take a look at [url=http://www.bungie.net/Inside/jobs.aspx#job13146]Bungie's Jobs page[/url], and the list of qualities and qualifications they look for in prospective employees. Taken individually, most of these stipulations are fairly familiar, especially if you've read as many recruitment adverts as I have. The relative emphasis is more revealing. Personal qualities - self-motivation, love of games, dedication to quality, willingness and ability to work collectively and harmoniously - are heavily emphasised and non-negotiable. Technical and experience requirements are relatively less specific, and there's clearly a willingness to be flexible, and consider individual candidates on their individual merits. The average HR department or recruitment agency uses the exact opposite of this pattern, and never thinks twice about it.

There's an important part of the answer. You hire people who are driven by their own professional pride and their love of games to do the best job possible, and you trust them to get on with the job, with a minimum of supervision. You turn away incorrigible prima-donnas, control-freaks, egotists and office-politicians. Telling the difference isn't at all easy - I was once responsible for hiring a full-blown industrial psychopath - but the results are definitely worth it. You don't get these results by hiring people with paper qualifications and standing over them with a whip. Taking a hint from Bungie's tantalising (or mischievous) nautical references, you could call these the submarine-crew and the galley-slave models of team management.

Putting a largely self-organising team together isn't the whole of the story, of course. You still have to deliver a finished, polished game on time, and here Bungie's record is inevitably less than perfect. I'd go so far as to say that this problem is quite simply insoluble. You can either have a finished, polished game, or you can have a game delivered to s strict deadline, but not both. The fact that a team with Bungie's collective skill, talent and experience can't reliably solve this problem should tell us something important, but denying this long-established fact ("Everything always takes longer and costs more") is an article of faith for the people wielding the whips in the galley-slave model. Their standard technique for dealing with the problem when it can no longer be ignored is to speed up the drumbeat, cram more slaves onto the rowing-benches and flog everybody harder. You can imagine how well that works, and the state of the crew when it's all over.

If you accept that this problem can't be adequately solved by deploying more copies of Microsoft Project and bigger sticks, then the only remaining way to mitigate it is to find a way to loosen up the deadline. This requires a serious effort to rethink and reshape the distribution channels. Valve's Steam is a laudable effort in this direction, though clearly not a perfect solution.

The other really valuable thing Bungie can offer to the industry is hope. I can't recall another instance where a games development house swallowed by a publisher has been able to win free again - though I must be fair, and acknowledge that Microsoft's handling of Bungie seems to have been exceptionally wise and far-sighted, by industry standards. I've experienced the swallowing process myself, and seen it happen to ex-colleagues who had managed to set up their own company. Bungie's example - and Microsoft's example - offer at least the faint possibility of escape to developers currently languishing in the bellies of their various beasts.

category: 
topic: 

Comments

I too was surprised by Bungie managing to go independent and break free (or should I say let go by?) Microsoft, and though I didn't know it then, yes, what I felt then was hope. The main reason being was that I was afraid that Microsoft was going to turn Bungie into a Halo Factory, and that would mean that after all these years of hoping(there's that word again) all chances of Bungie eventually develop Phoenix would be gone.

On the subject of steam, we've also seen a slight turn in that direction with console games as well, both with Xbox Live Originals over Xbox Live, Playsation Network, and the Wii Virtual Console. Even though these are all just ways to download older games, the technology could be adapted to current gen games with a more robust distribution network(which would obviously cost a truckload of money), but I can see the industry eventually migrating that way. I mean, how much do you thin it costs to download a game versus shipping it to various countries then various warehouses across miles and miles of road to finally get it to the store?

Maybe finally games wouldn't cost an arm and a leg, but knowing the business they would probably just keep the price the same and reap bigger profits.

-MAZ

PS: Is the fact that I only had to google one of the words in the title a good thing or a bad thing?

To AlexZander:

Thanks for replying - and increasing my incentive to avoid destructive hatchet-jobs in future.

Regarding the benefits of independence: yes, there were specific grounds for apprehension in the case of Bungie, Halo and Microsoft, but I stated (or over-stated) that paragraph more broadly because even the most benign of large companies are more prone to bureaucratic excess than small companies generally are, and (in case it wasn't already obvious) I regard over-management as the greatest obstacle to producing good games.

Regarding distribution: the examples you mention are encouraging, but all of these, including Steam, are proprietary, and most of them are platform-specific. As an alternative, there was an experiment I read about some years ago involving software being burnt to disc, and packaging and manuals printed, on demand in retail stores. I can imagine that burn times could be a problem, but there might be ways around that. As you point out, there's a sizeable amount of money which could be clawed back from the middle-men.

Regarding Atropos (I assume it was her you had to Google for): I'd say two out of three's not bad. Any hard-core Bungie-watcher needs to be able to cope with Classical references, and I must confess that I was only aware of Atropos through my early-teenage fascination with C.S. Forester's Hornblower novels - an H.M.S. Atropos was one of Hornblower's ships, and the description of the figurehead with the golden shears lodged in my mind. Besides, I found the notion of a Greek goddess of deadlines irresistible (yes, yes, I know she's not exactly a goddess).

Thanks for the writeup - and these comments are certainly applicable outside the gaming community as well. However as a longtime member of the software development community myself I have a few comments about some of the conclusions that you came to.

One point that you made is that there has to be a trade-off between dates and quality. The old 'Iron Triangle' actually states that there must be a trade-off between dates, features, and quality(serviceability). Recent shifts in software best practices dictate that all three can be improved without sacrificing anything (see Agile, Lean, Six Sigma). Though it's tough to peer into the black box that it Bungie, based on the job qualifications and the fact that they work in a large, open warehouse with no cubicles and frequently work in pairs or small teams, it appears that they're using some extreme programming (XP) best practices, which would allow them to maintain creative control and software quality while simultaneously limiting excessive documentation and defects.

The last thing that I think Bungie does extremely well is to constantly re-prioritize their product features and maintain a high-quality product all through their cycle. Long story short- they do important features first, then as those complete they slowly add new ones without breaking the first features. Then when the release date happens they don't have to rush to make the product work - they just cut the unimportant features. Features that almost got cut include online-coop and forge. Features that did get cut are a custom matchmaking browser, a few maps ideas, and a few known bug fixes.

In my mind, Bungie's example can be replicated across the software industry. It just requires that some old practices get re-evaluated. It's hard, but I've seen it done (and done it).

Anonymous wrote:

[quote]Thanks for the writeup - and these comments are certainly applicable outside the gaming community as well. However as a longtime member of the software development community myself I have a few comments about some of the conclusions that you came to.

One point that you made is that there has to be a trade-off between dates and quality. The old 'Iron Triangle' actually states that there must be a trade-off between dates, features, and quality(serviceability). Recent shifts in software best practices dictate that all three can be improved without sacrificing anything (see Agile, Lean, Six Sigma). Though it's tough to peer into the black box that it Bungie, based on the job qualifications and the fact that they work in a large, open warehouse with no cubicles and frequently work in pairs or small teams, it appears that they're using some extreme programming (XP) best practices, which would allow them to maintain creative control and software quality while simultaneously limiting excessive documentation and defects.

The last thing that I think Bungie does extremely well is to constantly re-prioritize their product features and maintain a high-quality product all through their cycle. Long story short- they do important features first, then as those complete they slowly add new ones without breaking the first features. Then when the release date happens they don't have to rush to make the product work - they just cut the unimportant features. Features that almost got cut include online-coop and forge. Features that did get cut are a custom matchmaking browser, a few maps ideas, and a few known bug fixes.

In my mind, Bungie's example can be replicated across the software industry. It just requires that some old practices get re-evaluated. It's hard, but I've seen it done (and done it).
[/quote]

Thanks for contributing. Your optimism is refreshing, and a worthwhile counterpoint to my embittered cynicism. Having said that, I think your many valid observations tend to reinforce, rather than weaken, my strongest point - that if Bungie can't reliably hit their deadlines AND achieve the degree of polish they strive for, who can?

Very nice. The make up of Bungies team has always been something interesting. It always seems like a very casual management structure. The emphasis on team moral and the way they try to eliminate the egotistical, political types can have pros and cons.

VVV wrote:

[quote]Very nice. The make up of Bungies team has always been something interesting. It always seems like a very casual management structure. The emphasis on team moral and the way they try to eliminate the egotistical, political types can have pros and cons.[/quote]

Thanks for the compliment. It might be superfluous to add that I have no privileged information regarding to Bungie's team dynamics, but the publicly-available evidence and my own experience of good and bad team-management strategies seemed to point to the same conclusions.

Your final sentence set me thinking, though. To me it's axiomatic that a heavily-managed team will never produce a truly great game, because that requires a willingness to experiment and take risks which is completely unthinkable in an organisation run on bureaucratic lines. I'm willing to admit the possibility that such a team might produce a competent game, but aiming for mediocrity and achieving it is nothing to get excited about. The likelihood of a bad game, though, seems rather high.

Completing the table: the best and worst projects I've ever worked on were under management regimes which could certainly be described as 'casual', though in very different senses of the word. Casual management is no guarantee of good performance, but I see it as the only environment in which a truly great game can be made. It's a classic risk/reward dilemma.

I should admit to a vested interest here; I'm a generalist (my enemies would probably call me a dilettante, if their vocabularies extended that far) who's never really happy wearing just one hat all the time. In the days of small, loosely-structured development teams I was a highly valuable asset. A highly-structured, heavily-managed team with narrowly-specialised job descriptions doesn't really have a slot for someone like me. And then there's my profound conviction that most bureaucratic activity is counter-productive, and that the predictive power of the scheduling systems I've seen in action makes tabloid horoscopes look good...

I've racked my brains to come up with plausible 'cons' to fit your last sentence, and I'm not doing well. If you have a technically-strong team with great esprit-de-corps who'd far rather get on with the job at hand than bicker and plot, then what's not to like? I am genuinely curious.

[quote=OldNick]I've racked my brains to come up with plausible 'cons' to fit your last sentence, and I'm not doing well. If you have a technically-strong team with great esprit-de-corps who'd far rather get on with the job at hand than bicker and plot, then what's not to like? I am genuinely curious.[/quote]
The only "con" I can come up with is passing over hiring a genius with the personality of a wounded badger in order to hire a person perhaps less competant in the field but who is less abrasive. And that's a fairly weak "con", IMO, as the loss of that particular spark of genius is a small price to pay to avoid having that same spark crushed in all who must deal with the SOB.

-- Steve

[quote=Anton P Nym]
The only "con" I can come up with is passing over hiring a genius with the personality of a wounded badger in order to hire a person perhaps less competant in the field but who is less abrasive. And that's a fairly weak "con", IMO, as the loss of that particular spark of genius is a small price to pay to avoid having that same spark crushed in all who must deal with the SOB.

-- Steve[/quote]

Heartily seconded, even though some of my ex-colleagues would probably think that the "genius with the personality of a wounded badger" description fits me all too well. All I can do is keep taking the medication and wearing the muzzle...

Nice reply to my comment Nick. In business I have seen very rigid management structures as well as "casual" ones. The Casual structure certainly has more going for it from my personal point of view for all the reason you have described. However I have seen some things lacking. The problem is I'm in a very different field of work and therefore it is probably something I can't relate to the production of a video game or the like.

What may be factors is things like too many people leaving the direction the majority of the people involved set out upon. Meh, maybe.

This may come in many forms. Plot, time table or character development. Without a strict direction and with encouragement to be independent all sorts of things [i]could[/i] happen.