GDC 2005

1
(No Transcript)
2
Small Teams Big Dreams How to Make an MMORPG
without a Massive Team or Budget
NetDevil Scott Brown, CEO Hermann
Peterscheck, Producer
3

• Basic Overview
• MMOs have changed over the years.
• Started Small
• MUDs would peak at hundreds of users, mostly
  free.
• Ultima Online had about 250k, was expected to
  have 10-20k.
• Became Huge Fast
• Over 1 billion by 2006 in the West.
• Estimated to be 1.5 billion by 2011.
• Not Well Distributed
• World of Warcraft was about 471m of this.

4

• Basic Overview
• Budgets Grew as Well
• Started largely as independent university
  projects.
• Became experiments at professional developers,
  later becoming more massive projects.
• Have now grown to among the most expensive game
  development projects with 40m-60m becoming more
  common.

5

• VooDoo Predicting the Future
• Weve seen these charts lately

6

• VooDoo Predicting the Future
• Television An Analog. of Channels Skyrocket.

7

• VooDoo Predicting the Future
• Television An Analog. Revenue Flat after 1998.

8

• VooDoo Predicting the Future
• Ads Games Huge Money in the Future?

9

• Another Look at Growth.
• Looks like massive growth

10

• Another Look at Growth.
• Without WoW its pretty flat

11

• So What?
• Levels of success perhaps exaggerated.
• How do we know this?
• Personal experience.
• High-Profile failures.
• Cancelled Projects.
• Uneven distribution of success.
• The top few games have a huge majority of the
  users, and that is the way most entertainment is.

12

• So What? Vicious Cycle.
• Investors need to believe rapid growth to justify
  investment.

13

• So What? Vicious Cycle.
• Developers need to believe to get a budget to
  compete with all the other people with 40
  million

14

• So What? Vicious Cycle.
• Researchers and press people need to write
  interesting papers that have BIG NUMBERS!

15

• So What? Vicious Cycle.
• And then theres funny charts ?.

16

• It starts with the deal.
• Build a prototype first.
• Hard to get MMO deal without something to see.
• Publishers perceive too much risk.
• Focus on visuals and performance quality.
• When people see something pretty that runs well,
  they are impressed.

17

• It starts with the deal.
• Deal Considerations.
• Light on specific milestone definitions.
• At the beginning of the project you know less
  than at any other point.
• How can you create a detailed milestone document
  at this point?
• Make sure both parties interests are aligned.
• Focus on keeping costs low and development
  efficient.
• Honestly evaluate the market viability of the
  product.
• Keep teams size as small as possible for as long
  as possible.

18

• Corporate Structure
• Low Fat
• Make most of your company development, not
  administration.
• Outsource everything you can (payroll, taxes,
  legal, finance planning, hr, recruiting).
• Find cheap rent.
• NetDevil went from a small office in a bank
  building to an industrial office park.
• Avoided 15th floor penthouse in downtown area.
• Think of the cost difference between 14/sf and
  75/sf when you have 10,000 sqft.

19

• Corporate Structure
• Quality of Life
• Crunching is not a good thing for you team.
• Happy rested employees are more productive.
• Use as a absolute last resort, you should be able
  to get things done without constant crunching.
• Focus on the Products
• as fun as little things like rec centers, cool
  furniture, etc. can be for an office, the best
  way to have happy employees is to work on great
  projects the team believes in.

20

• Small Project Overview
• Lower spec Larger user base.
• Think of all those people that bought the Dell
  computer with 512 MB of RAM and an integrated
  video card.
• Can you afford to not support them?
• Stylized vs. Real.
• This doesnt mean cartoony.
• If you go for real, the next real game will
  make yours look dated.
• Especially bad for MMOs with long tails.

21

• Small Project Overview

22

• Small Development Teams in Reality
• Why are MMOs so damn expensive!?!?
• Lack of Knowledge of What MMOs require.
• Unusual skills.
• Client/Server Architecture.
• Complex database issues.
• Always have to consider LOTS of people doing
  something.
• Cheating and Exploits.
• Example of increased complexity.
• Inventory on single player game.
• Move object from one location to another.
• Inventory on MMO.
• Send request to do transaction from client to
  server.
• Check validity of request on server.
• Process request on server.
• Send response to client.
• Update client based on response.
• Trivial tasks become very nasty!

23

• Small Development Teams in Reality
• Why are MMOs so damn expensive!?!?
• Tons of Content.
• 10,000 quests.
• 20 people working
• 2080 work hours in a year.
• About 1 quest every 4 hours.
• 10,000 items
• 10 people, 2080 hours/year.
• About 1 item every 2 hours.
• 50 unique areas.
• 10 people, 2080 hours/year.
• 416 hours/area. (just over 10 days).
• With that we have a 40 person content team and
  its tight!
• Cost of 40 person content team/year approx.
  400,000/mo 12 4,800,000/year.
• Wheres the testing, balancing, bug fixing, etc?

24

• Small Development Teams in Reality
• Why are MMOs so damn expensive!?!?
• A Product and A Service.
• Billing. Potential startup cost
• A billing can cost anywhere from 400 to
  400,000.
• Off the shelf solutions.
• Custom built applications.
• Billing will represent 2.5 - 7 of your revenue.
• Infrastructure for launch.
• Expect 100k customers.
• Expect 40-60 simultaneous day 1 users.
• Expect 3000 subs per shard.
• Need 20 shards.
• Shards can cost 5,000 to 250,000 depending on
  how you build them.
• AA was 20 computers (blade) SQL license. Approx
  100K/shard .
• Jumpate Shard appros 3,500/shard / month.
• Finding a partner that does monthly and can scale
  as you need is a great way to control costs
• Not suggesting one is better than the other,
  just know your costs early.
• Bandwidth scales similarly.

25

• Small Development Teams in Reality
• Why are MMOs so damn expensive!?!?
• A Product and A Service.
• Technical Support.
• A trouble ticket system can cost 5,000-50,000
  monthly fees.
• Tier 1 and Tier 2 technical support staff.
• Customer Service.
• Big MMOs can require a lot of support!
• World of Warcraft is said to have 2000 customer
  service representatives.
• About 1 Rep/3000 people. (although this doesnt
  scale linearly)
• If you expect 100k subs you ramp up 30 CSRs.

26

• Small Development Teams in Reality
• Why are MMOs so damn expensive!?!?
• Summary The cost in addition to development.
• Content Development 5milion - 15million (40
  people 1-3 years)
• Billing System 500,000
• Customer Service 300,000/mo ramp up costs.
• Technical Support 150,000/mo ramp up costs.
• If you ramp up too early, your monthly burn will
  really hurt.
• If you guess the number of day one players
  incorrectly, you will bleed money fast.

27

• Small Development Teams in Reality
• Some other scenarios.

28

• Small Development Teams in Reality
• Why are MMOs so damn expensive!?!?
• MMOs are notorious for slipping.
• Slipping happens at the end, after infrastructure
  is set up and people are hired.
• Cost of slipping 1 year.
• 300k/mo in support and fees.
• 400k/mo in content development.
• 250k/mo in game development.
• Slip Late 11.4 Million/year
• Slip Early 9 Million/year
• Slip Early, less content 4.8 Million/year

29

• Small Development Teams in Reality
• How is a small team possible?
• Reduce the content you need.
• Grow the service as necessary.
• Dont take on the giant games head on.
• You cant build World of Warcraft or Lord of
  the Rings Online with 4 million.
• You cant build a content heavy, custom mission
  based game with a small team.
• You cant build a game that requires 40 in house
  testers for 2 years to balance for 4 million.
• You cant build a OMG-revolutionary-never-seen-bef
  ore MMO cheaply (and well).

30

• Small Development Teams in Reality
• How is a small team possible?
• Case Study Club Penguin.
• In March of 2005, New Horizon Interactive set
  out to create an online world for kids where they
  could safely play games, have fun and interact.
• Eight whirlwind months later, in October 2005,
  Club Penguin opened to the public with about
  25,000 users.
• From ClubPenguin.com website.
• Club Penguin is not World of Warcraft.
• 8 months, 25k subs is much more profitable than
  many 10-40 million games.

31

• Small Development Teams in Reality
• How is a small team possible?
• Core Directives.
• Entertainment value is not proportional to
  development cost.
• A little greatness is always better than a lot of
  mediocrity.
• You must externally measure the acceptance of
  your game.
• Get to the core game as fast as possible.
• Keep things small and focused.
• Design for maximum player fun hour per
  product/feature development hour.
• Example Custom Missions 40-80 dev hours per 1
  gameplay hour.
• Example Rep Grinding 100-200 dev hour per
  200-400 gameplay hours.

32

• Small Development Teams in Reality
• Get to know your friends!
• Scrum and Agile.
• Rapid Prototyping.
• Thin or Vertical Slicing.
• Test based development.
• Get to identify your foes!
• Feature Creep.
• Slipping late instead of slipping early.
• Confusing RD with Production.
• More People does not Shorter Development at the
  RD phase.
• In terms of content it can make a big difference.
• In terms of RD it is less so.
• In terms of post RD feature development more
  people can shorten time.

33

• Jumpgate Evolution Rapid Prototyping
• Team of 5 people
• Concept Artist
• 1 Artists
• 2 Programmers
• Producer
• Time and Cost 3 Months - 150,000.
• Rush to graphics look and feel FIRST.
• Counterintuitive.
• Graphic limitations drive design limitations.
• Goal was to run on diverse hardware and look
  great.
• Always make sure it works on the minimum spec.

34

• Jumpgate Evolution Rapid Prototyping
• Start with great concept art.
• Iterate at the concept phase as much as possible
  to ensure highest quality.
• Implement the look and feel as quickly as
  possible.
• Dont focus on fancy tools, solid pipelines at
  this point.
• Get it looking good and running well.
• Maintain performance and reliability.
• Define the limitations of the project early.
• Number of polys, texture resolution.
• How will you be seeing things?
• Small or Big, up close or far away.
• Our Target
• 512 MB of Ram
• 64 MB non-shader cards

35

• Jumpgate Evolution Rapid Prototyping

36

• Jumpgate Evolution Rapid Prototyping

37

• Jumpgate Evolution Rapid Prototyping

38

• Jumpgate Evolution Rapid Prototyping

39

• Jumpgate Evolution Rapid Prototyping

40

• Jumpgate Evolution Rapid Prototyping
• Simultaneously build basic gameplay.
• Flying around in space.
• Shooting.
• Blowing things up.
• Avoiding obstacles.
• Thats about it!
• If the basic activity is not fun, the game will
  never be fun.
• Basic activity stuff that people will be doing
  a majority of the time.

41

• Jumpgate Evolution Rapid Prototyping
• Autobuild/Continuous Integration system.
• This was critical to keeping the code clean and
  assets up to date.
• Develop in a live environment as soon as
  possible!
• Team members to check in high quality work
  instead of waiting for QA to find problems.
• Requires that builds take less time (5-15
  minutes) and happen automatically.
• name the tools we use to build this for specifics
• Prevents build oddities.
• Less crashes.
• No weird rules for builds
• (ie make sure you change settings x in .ini file
  in this version, are all the art files checked
  in, etc).
• Force priority of crashes and performance issues.
• Stability and performance need to always be top
  prioty.
• Never put either of these off for later thinking
  but so much will change its a waste to fix it
  now.
• There is no later.

42

• Jumpgate Evolution Rapid Prototyping
• More Important What we did NOT do.
• No fancy editors.
• No complex, well defined art pipeline (YET!)
• No super detailed design document.
• No massive, complex database, login server, etc.
• All of this was quick and dirty.
• Remember the goal An online game with the most
  basic elements required to prove the look and
  feel with simple gameplay.

43

• Jumpgate Evolution Rapid Prototyping
• Landmines.
• Good generalists instead of specialists.
• Concept artist might have to do basic texturing
  or UI.
• Graphics programmer may have to write some
  network code.
• EVERYONE has to be a designer to one degree or
  another.
• Resist the wouldnt it be cool if scenario.
• Resist the what about the questions.
• Those come later.
• No really, they dont matter now, they come
  later.

44

• Jumpgate Evolution Thin/Vertical Slice
• Thin Slice Phase 10 People/9 months (approx
  cost 900,000)
• 6 Programmers
• 3 Artists
• 1 Producer
• External Writing resource
• External Audio Resource

45

• Jumpgate Evolution Thin/Vertical Slice
• General Goals
• Build features in order of importance.
• Build content that can be reused or built
  quickly.
• Maintain performance and quality.
• Stay on a feature until it is done (i.e.
  shippable).
• Test and iterate early and often.
• Eat your own dog food. Play every day or two.
• Focus on what's really important.
• People fix things they might otherwise ignore.
• If you dont think its fun, nobody else will.
• Dangers
• Choosing 3rd party tools.
• Especially ones that are in no shipped products,
  this has almost 100 been a failure case for us
  over the last 10 years.
• Not testing enough.
• Underestimating number of iterations.
• Underestimating time to create content.

46

• Jumpgate Evolution Thin/Vertical Slice
• Dyanmic Content
• Missions are dynamically generated.
• We built a scenario generator.
• Make missions pseudo-random.
• Be able to reuse specific mission text within
  multiple scenarios.
• Reuse of art assets.
• Built ships so that they can be morphed and
  re-textured.
• Creates lots of variety without having to concept
  and model each individual asset.

47

• Jumpgate Evolution Thin/Vertical Slice
• Focus on a few great features not everything to
  everyone.
• Think about games with less content and high
  replayability.
• Diablo 2
• A few missions, tons of items.
• Random maps and loot, heavy reuse of creature and
  building assets.
• Gauntlet
• A few items and creatures, tons of levels.
• Elite
• Massively procedurally generated world with
  little custom content.
• Even World of Warcraft reuses terrain, buildings,
  creatures, icons, etc.

48

• Jumpgate Evolution Thin/Vertical Slice
• Focus on a few great features not everything to
  everyone.
• Whats fun about space games?
• Ship to ship combat.
• Ship upgrading.
• Missiles and guns.
• Large scale battles.

49

• Jumpgate Evolution Thin/Vertical Slice
• Focus on a few great features not everything to
  everyone.
• Secondary elements.
• Mining and trading.
• Manufacturing (crafting).
• Multiple progression.
• Rank (Level)
• Licenses.
• Faction rating as unlocking tool.
• Medals and Titles
• PvP
• Instanced and Public.
• PvP rewards similar to PvE rewards.
• Other people make great content.

50

• Jumpgate Evolution Thin/Vertical Slice
• Focus on a few great features not everything to
  everyone.
• Avoid the temptation to must have everything!
• City of Heroes launched with no loot or auction
  house.
• World of Warcraft does not have housing.
• Club Penguin is missing most core MMO features.
• All of these games have at least one GREAT
  element and a few secondary elements.
• What is yours?
• Focus on it.
• 3 great features gt 30 mediocre features.

51

• Jumpgate Evolution Thin/Vertical Slice
• Test Early Test Often
• Start with core feature.
• For Jumpgate Evolution it is space flight/combat.
• We had over 50 iterations on the first combat
  experience before it passed play testing.
• Focus testing. Requires discipline!
• Only give the person login/password information.
• Encourage the tester to comment.
• NEVER answer questions.
• Force team members to watch, but never say
  anything.
• Follow up with questions pertinent to game.
• After 15 minutes tell them they can stop if they
  want.
• If they stop, its a bad sign if they keep
  wanting to play you are on the right path!
• People will complain but do they keep playing?
• Different players actually all tend to get stuck
  in exactly the same way.
• Great way to resolve differences of opinions on
  team, its not about what team thinks, its about
  what players see and do. If the player doesnt
  understand who cares if it seems obvious?

52

• Jumpgate Evolution Thin/Vertical Slice
• Test Early Test Often
• Specific examples.
• Example Issue Most testers never saw the
  launch button and got stuck.
• Solution We had to make the launch button pulse
  and glow green.
• Example Issue We had a huge red reticule to
  denote targets. 8/10 people never saw it.
• Solution until we made it bigger and brighter.
• Example Issue We had a tester ask for text in
  the middle of the screen telling her about the
  mission status. At the moment she said it, there
  was text on the screen with that indication.
• Solution It turned out to be a problem that the
  explosion was obscuring the text long enough that
  she didnt see it.

53

• Jumpgate Evolution Thin/Vertical Slice
• Test Early Test Often
• Dont ignore problems!
• Dedication to test driven development means
  dedicating resources to it.
• We had about 20-40 of our development time on
  these little things.
• Things that make people quit that you wouldnt
  think would make them quit
• Mouse ship controls are a bit jerky.
• Periodic hitches in frame rate.
• Small delays in interface response.
• We had ALL of these.
• Score your game based on the testers willingness
  to recommend the game to a friend.
• Example Our score dropped from 7-9/10 to 4-5/10
  when we added an intro flythrough but didnt add
  voice-over yet.

54

• Jumpgate Evolution Thin/Vertical Slice
• Scrum and Agile in a nutshell.
• Scrum is your friend.
• We used 2 weeks sprints.
• Gives people focus.
• Makes people work as a team.
• We forced public demos.
• Always added a bit of extra polish.
• No one wants to look bad in front of their peers.
• Think about the game as someone else will
  experience it, which is the whole point.
• Dont use it as a reward/punishment system.
• Hard to be meticulous about maintaining the demos
  and the daily meetings.
• Push the game from dev to live (or QA) at the
  end of each sprint.
• Keep the we are live mentality.

55

• Jumpgate Evolution Avoiding Bad Decisions
• Avoid poorly evaluated middleware.
• People like to believe that by buying middleware
  dev time will be cut in half.
• Avoid feature creep.
• Do you really need that feature?
• No really?
• Delay early not late.
• Remember the graph of financial devastation.
• Do not expect that the next version of
  something will necessarily be faster.
• Always think about dev hour per entertainment
  hour. Its really cool isnt always the best
  reason to do something.
• Remember the audience.
• If you dont know who it is, is there one?
• Drink your Kool-aid.
• If you dont play your game, why should anyone
  else?

56

• Jumpgate Evolution Avoiding Bad Decisions
• Listen to your team.
• If your small team of developers dont believe in
  the product how will they make it great?
• Avoid the valley of delusion.
• Just because something is getting better does
  not mean it is great.
• As time gets short and money becomes tight,
  people will believe what they want to believe not
  what is so.
• Hint Use outside testers.
• Hire generalists
• What is that 3d texture artist going to do when
  the concept pipeline is clogged.
• Can he/she do UI? Special FX?
• What is that awesome network programmer going to
  do when all your bugs are UI related?
• Can/Will he/she learn UI or graphics code?
• Beware of the burnout monster it is slow and
  stealthy.
• People on small teams will often deliberately
  overwork themselves.
• Force them not to.
• 100 hours/week is not the same as 2.5 40 hour
  weeks.
• Save crunching for when you really have to.

57

• Jumpgate Evolution Summary
• Its possible to make an MMO with a small team
  (8-15 people) for less money (2-4 million).
• Do it with less!
• Why try and make the next World of Warcraft?
• Lots of companies are already trying this.
• Do you have their money and experience?
• Even with it, can you really meet or beat them?
• Focus on a core game to an underserved market.
• Is it really that fun to do what everyone else
  does anyway?
• Focus on the strengths of the game.
• If you cant make it great, dont make it at all.
• Less is more.
• Remember 3 great features gt 30 mediocre features!
• You get a lot more chances!
• If it costs 4 million, you get 10 tries for each
  40 million project.
• Itll be easier to find a publisher.

58

• Questions?