Presentation at GDC 2015 by Creative Producer Jean Pierre Kellams at PlatinumGames and Senior Coach Jon Leslie at Hansoft. Read more: ow.ly/JwCWc
PlatinumGames and Hansoft - the Road to Agility
More and more teams are moving towards agile or hybrids of agile and traditional methods. Managing the transition can be a challenge and it is important to align your teams, use the right metrics and organize the work in a good way in order to get studio wide buy-in and a satisfying result. In this session, Jean Pierre Kellams, Creative Producer of Scalebound from Osaka, Japan-based developer PlatinumGames, creator of titles such as Bayonetta, and Jon Leslie from Hansoft will tell you how the team at PlatinumGames made the switch. They will take you through the process, starting with the strengths and weaknesses that brought about the change, the actual transition, and the successes and failures that led to the realization that what seems ideal isn't always the case.
Machine Learning Software Engineering Patterns and Their Engineering
PlatinumGames Road to Agility with Hansoft
1. PlatinumGames and Hansoft:
The Road to Agility
By Jean Pierre Kellams, Creative Producer at PlatinumGames
with Jon Leslie, Senior Coach at Hansoft
@platinumgames #GDC15 @hansoft
15. WHAT MAKES HANSOFT DIFFERENT
COLLABORATION
Tasks, newsfeed, chat &
document sharing
Jonathan Ellis just now
Fix zooming bug
èCompleted
Raechel Lambert 5 mins ago
Update release notes
èCompleted
Salil Agarwal 8 mins ago
Schedule client meeting
èCompleted
News feed
16. WHAT MAKES HANSOFT DIFFERENT
RELENTLESS
TRANSPARENCY
Everyone sees
everything
17. WHAT MAKES HANSOFT DIFFERENT
MIXED
METHODS
Agile, Scrum, Kanban,
Gantt & Waterfall
18. WHAT MAKES HANSOFT DIFFERENT
DASHBOARDS &
BUSINESS
INTELLIGENCE
Actionable Agile
Metrics
20. Learning Curves
Applying a methodology and transitioning a team to
AGILE-like practices at PlatinumGames.
21. Who Am I?
Creative Producer on Scalebound
• First project as a lead producer
• I’ve touched every PG developed game in some capacity,
whether it be localization, music supervision, or writing.
The first “foreign” employee of PG
• Worked with the team when everyone was at Clover.
• Have developed a trust relationship.
A classic “Type A” personality:
• Volunteered for this gig (Dumbest move ever!) .
• Want desperately to make a difference and make this team
the strongest they can be.
22. An expert on production:
• I haven’t shipped anything as a lead, so I may be entirely full of it.
Feel free to think that.
• That being said, in my previous life I’ve worked with more producers, creative
directors, and teams than most do in a lifetime. I hope that I’ve learned the good
and forgotten the bad.
An expert on AGILE:
• If I was, I’d be Jon.
• I’m learning as I go, and I am learning that experts in methodologies often place
the methodology before the result. Maybe that is good, maybe that is bad.
• Ask me in a year. ☺
Who I Am Not
23. • We’ve been using Hansoft for all of our projects, since the end
of The Wonderful 101.
• Bayonetta 2 used Hansoft to “push to ship.”
• This was a Gantt/Waterfall model effort to execute on the
final parts of the game.
• Projects currently in production are using Hansoft to help shift
to AGILE methodology. Today’s talk is my story in helping drive
this progress.
PlatinumGames + Hansoft
24. Discovery through iteration.
• Quality comes from learnings.
Generalists, not specialists.
• Our team are very adaptable.
Making the right decisions for the game.
• The game always comes first, even to a fault.
Understanding What Makes Us Strong…
25. • Lack of documentation.
• Open floor plan and ad-hoc decision making can make information sharing hard.
• Iteration management.
• We almost never ship the first things we make.
• Fear of future iterations can force waiting, which degrades the ability to learn.
• Lack of methodology hurts our ability to schedule precisely.
• Ad-hoc scoping/backlogs.
• Introduces some fuzziness to goal setting.
…And What Makes Us Weak
26. What Do We Do
• What we already do:
• Granular goals and decision making based on iteration.
• What makes us stronger?
• Iterations.
• Committing just in time.
• Embracing the unknown.
• How do we overcome weaknesses?
• Making goals clearer.
• Using methodology to guide decision making.
• Tools to track goals that match this methodology.
• Tracking capacity and iteration.
27. • Early research into tools and methodologies:
• MS Project – Waterfall/Gantt is incompatible with how we make
games.
• Redmine – Good way to trigger backlog tasks or track bugs, not
really good for other things.
• Hansoft – Goldilocks software?
Multiple methodologies, but tuned for AGILE.
Introducing a Methodology
28. Adopt a standard methodology,
find a tool that uses that
methodology, iterate on our usage
of that methodology until it
incorporates our best practices.
The Solution
29. Early Project Milestone A
Early Project Milestone B
Early Project Milestone C
Early Trial And Error Was Key
30. • What’s going on?
• This milestone only carried work
from our contract. The burndown
looks great but since we weren’t used
to the software it only captured a
small portion of the work we were
doing.
• How did we solve this?
• We decided to load up all the work
into a milestone.
Learning Curves
31. • Try, try again.
• We switched to loading everything into milestone. The curve looks great, but this
was a less than successful sprint, because our sprint duration was too long and
we began dumping work off to another sprint.
• How did we solve this?
• We needed to change how we approached sprints.
Learning Curves
32. • Try, try again. Again.
• Our curve here is really flat because we decided to let our team add and remove
work during the course of a sprint, instead of just killing the work like we did in the
previous milestone.
• How did we solve this?
• We decided we needed more help.
Learning Curves
33. • We realized we didn’t want to keep working this way,
so we sought advice.
• Hansoft coaches are great places to go for this.
• Explain where you think you are failing and they can help you to solve that.
• The advice we got:
• Structure shorter sprints.
• Look at who is working together and turn them into functional groups.
• Let them name themselves and take responsibility for their work.
• Gantt is OK for some sections.
Engaging Hansoft Coaches
35. • We put the team on a weekly sprint cadence with clearer goals.
• Weekly build checks enforced responsibility and tightened feedback loops.
• Build testing prior to the checks increased build stability.
• The team was now predictable.
• We were able to predict success on individual sprints.
• Individual team members had a much better idea of how much work they could
handle.
• But we still didn’t know how much work the team could handle!
Putting Advice Into Action
37. We had “backlog style schedules”.
• Design would schedule what they’d like to think about when.
We had “allocation style schedules”.
• This environment artist is doing this task this week.
We had hybrid allocation/backlogs schedules.
• Sound would say that they need to be doing all this, but I’m not sure if we will, or
can.
We had Gantt schedules.
• Modelers would say you can expect my section to do this at this time.
Improving At Scheduling/Backlogs
38. • We had “backlog style schedules”.
• Design would schedule what they’d like to think about when.
• We had “allocation style schedules”.
• This environment artist is doing this task this week.
• We had hybrid allocation/backlogs schedules.
• Sound would say that they need to be doing all this, but I’m not sure if we will, or
can.
• We had Gantt schedules.
• Modelers would say you can expect my section to do this at this time.
And they were all
made in Excel...
With WordArt!
With
Word Art!
To Give You An Idea:
39. • We realized that we needed to create a backlog and try to bring
everything into view. We had a semi-clear roadmap for doing this:
• Standardizing the format.
• Prioritization discussions.
• Migrating the format to Hansoft.
• ...
• PROFIT!
The Road To A Reasonable Backlog…
40. • Using Excel was a conscious choice.
• Comfortable software to bring the
team around to a new idea.
• Mimicking some of the Hansoft
functionality in the sheet so the
team understands the information
they need to prioritize.
• “Gantt”-style visualization to help
the more visually oriented leads
figure out how all the puzzle pieces
fit.
Standardizing the Format
41. • Allowing for some of the extraneous information from previous
scheduling formats, but structuring them in a way that it would help
future migrations to the software also pushed new thinking.
• However, we were still too reliant on average estimates and not relative
estimates/points.
• Prioritization and scoping discussions were key to both starting to
determine relative size and milestone planning.
This Was A Good Baby Step
42. • Designers at Platinum HATE cutting anything.
• Using the word cut might end with you being cut. ☺
• But a designer’s “backlog” isn’t by definition reasonable.
• Some tricks to trick your team into cutting early, scoping appropriately,
and not ending up being the producer who threw water on a creative
fire:
• “What do you want to see the most?”
• “What do you want us to make first?”
• “Compared to X, is this bigger? Smaller? About the same?”
• “Can this X really by Y+Z instead?”
PROTIP: How to Talk About Priorities…
…Without Talking About Priorities
43. This predictability, brought about by iterating on our process with the help of
Hansoft coaches and using the software to visualize our understandings has
led to deeper understanding of development by the team and a desire to track
their work, and more importantly their progress, using the software.
How Did We End Up? Crazy Predictable
44. • Average hours for the same stage in other projects are way down.
• Team direction and approach is easier to explain.
• Schedule discussions have a tone of moving forward,
not the fear we had prior.
Profit!? Maybe!
45. • Milestones are still mini-backlogs. We could have a much stronger plan/
master backlog.
• Transitioning to this now, but also aware of the right way to start the next project.
• Our learnings are influencing new projects at PG to start with strong backlogs.
• If we can get them to understand backlogs and velocity, prioritization will be easy in the
future.
Where We Can Still Improve
46. • Hansoft is a tool. You are the creator.
• The ideal curve is not always ideal.
• Theory and methodology are secondary to results for your team.
• You make decisions, data guides those decisions, methodology provides the
framework for the decision. But you are the key.
• Your team’s experience is your most valuable asset. Trust them before you trust a
tool. The tool is there to help them.
• Find places for constant improvement.
• The better you make things, the better you serve your team by creating an
environment for them to be the amazing creators they are.
• Baby steps are the best steps.
What’s Your Point, Kid?
47. We Are Hiring!
WE ARE HIRING!
We are making incredible things that we can’t wait to talk
about. We’d like to make them with you.
You’ll love Japan. Trust me.