GeekStack Postmortem

Posted on | August 19, 2011 | 7 Comments

GeekStack is officially dead, for now.  I haven’t worked on it in forever and it’s no longer the driving force in my mind so there is no resurgence to hope for.  (Fans, please read to the end.  There is a little bright spot.)  I need to get this off my chest so I can move on.

A brief history: I got the idea for a trading card game featuring the heroes of science and technology back in the Fall of 2008.  I did some planning and research for a few months, and in spring and summer of 2009 I did a lot of development for it.  In fall of 2009 I got a cofounder and that worked for about 6 months until that relationship failed slowly and silently.  My work slowed to a halt until summer of 2010 when it ceased.  Now, over a year later, I’m going to go over the reasons why I failed and what I can do about  that in future ventures.  Hopefully you can learn something from this, but this is mostly for my own benefit.  Here’s a list of some of the things I did wrong.

1. I Was Planning to Launch a Mature Company

Good grief, you should have seen some of the things I spent time and energy working on.  Partnerships with other companies.  Building a generalized game engine that could accomodate any known trading card game.  Plans for community participation in the sourcing of future characters.  Many hundreds of scientist bios sorted by category.  Reading multiple books on game design.  Structuring online tournaments.  A release schedule for themed sets of cars going years into the future.  In my head, I was running the company that would have resulted with ~2-5 years of perfect execution.  Silly me, I glossed over the work necessary to get to that point.

2. My MVP Was WAAAAAAAAAAY Too Complex

Like a good boy, I did some market research into CCGs in general, and ran surveys for my specific idea.  Emboldened by good results, I started working on my MVP of … a completely generalized game engine running on a realtime gameplay platform.  I wrote a pretty good game engine, but it consumed 100% of my development time (see also #5), but the realtime part was beyond my grasp (see also #2a)

I should have gone roughly in this order:

1. you can collect and view cards (with no game rules)
2. cards get one new rule
3. you can play the game with 100% page refreshes every time you do anything, including having to guess and check when your opponent is finished with their turn (imagine chess by mail)
4. when customers can successfully use that feature, return to step 2
5. once the gameplay is fun, work on player experience features (drag and drop cards, async updates to the page, etc)

Why didn’t I do this?  First, #3 would mean that it would super duper suck to try and play the game, and I didn’t want that to be the first impression. Second, I was trying to optimize for my personal development flow instead of product development flow.  The goal was quicker total development time, but it instead led to burnout and quitting due to lack of visible progress.  Finally, I had in mind the “charge from Day 1″ mantra and I knew that it would be a long time before this gave anything worth paying for.  So I basically waterfalled myself to death.

2a. My MVP Was Beyond My Ability

Making an MVP is not the place to be a hero – it’s an feedback and verification phase.  Part of my planned MVP was a real-time multiplayer card playing surface.  Remember, this was 2009, so no Node.js, a much less mature EventMachine, etc.  (If I get any dates wrong, remember this is my perception of what was out there.  People were doing this stuff, just not me.)

There were two big problems: not only was this extremely overkill as an MVP, it was beyond my technical abilities.  I did not have the skill to prototype this.  My plan was literally a) learn EventMachine and use that, b) learn Python and use Tornado (I heard it was good), or c) learn Erlang because that was the “right way” to do event driven server programming.  I was not confident enough in any of those choices – I figured EventMachine was the best idea given that I was writing the site in Rails but I couldn’t find documentation and tutorials enough to give me the courage to actually work on it (also next point as well).  I was aware enough of the deployment and integration difficulties of managing multiple platforms to know that I probably wasn’t going to use Python or Erlang, but that just helped me not choose those options.

I think if I knew in real life someone that had been using EventMachine and could give me a 10 minute overview and answer basic questions like “How do I pass a user between a Rails app and EM evented code?”, that probably would have been enough (alternately, I could have reached forward through time to get the PeepCode episodes on EventMachine.  Instead, I just left it as unknown and went back to work I knew how to do because I had no priorities or deadlines to force the issue of how to code the gameplay part of the app.

3. Analysis Paralysis for Unnecessary Topics

Speaking of time, I continually spent 10+ hours researching questions that could have been answered in 2 minutes by a knowledgeable mentor.  These were things that wouldn’t make or break my business, but I was so worried about wasting time later to fix something I did incorrectly at the beginning that I just ended up researching more or punting.  Things like corporate structure, how to collect payments, where to get a merchant account, which collaboration and project planning software to use, etc.  I did plenty of research on the Internet, and found plenty of facts, but there was no one to tell me “Given my knowledge and your situation, you should do this, this, and this”.  I believe I ended up making correct decisions, but it took massive amounts of clock and calendar time for me to feel confident about them.

4. Unrealistic Financial Expectations Clouded My Decision Making

A big part of the reason I had become interested in startups was because of Paul Graham’s writing.  Essays like How To Make Wealth made so much sense.  I never wanted to be a billionaire or world famous so the Bill Gates and Steve Jobs stories didn’t appeal to me, but the idea of solving “the money problem” did.

By this time, I had two young kids and I wanted to spend time with my family, not at work.  But because of said family, I had higher expenses I didn’t have unilateral power to decide what I did or how we lived.  We had some savings but also leftover student loans, a mortgage, etc so my runway was helicopter short.  Quitting the day job was not an option, so the target became *replacing* the day job.  This meant ~$10K per month to replace salary, pay COBRA, etc.  This was aside from expenses that the business needed to run, biggest of which was a whole bunch of art assets that I couldn’t produce.

It may seem obvious that it’s really hard to get from $0/month to $10,000/month, but that was my goal.  I made decisions based on that premise – will this help me earn $10K/mo?  I ended up doing a micro version of the VC shoot-for-the-moon game, because I was unwilling to consider a plan that led to making $250/month and included enough momentum to keep going next month.  It was all or nothing, and nothing won.

5. I Did Things in the Wrong Order

As part of my way-too-huge plan, I figured out that I needed four things: a flexible game engine, a website that included basic administration and the ability to play the game, a game design including ~200 cards, and a community of ~50 playtesters.  I had 10-15 hours a week, so I figured that this would take me years to do all those things.  I worried that I wouldn’t make progress if if I tried to do them all at once, so I tried to do them sequentially.  I managed to finish the game engine and get a crude version of the administrative website done.

The problem with this is that even though I made good progress on those parts, I made zero progress on the other parts during that time.  Extrapolating from 0% progress over 6 months  is pretty discouraging.

6. I Was in Denial About Dependencies

The schedule problems from #5 also meant that it was impossible for me to work with others because I had no ability to schedule dependencies.  For instance, I knew early that I wanted to feature the real life heroes of science and technology, and so I went to the TechCrunch50 conference, where I met some awesome people that were interested in being on the cards.  I tried following up with them on how that would work, but because I was being frugal about how and when I spent money, I didn’t want to commission the illustrations of them.  So I just ended up coming off as a flake because I didn’t follow through.  This made me even more hesitant to work with people in the future, so it ended up being either done by me or punted indefinitely.

By the later stages of my winding down, I knew that I needed another partner, either an artist so I could save cash and get some design help, or a developer to push functionality faster.  But I had been moving so slowly on my own that I had two worries: that the search for a partner would take away the rest of the little time I was spending, and that if I did find a partner, they would quickly be disillusioned by how little work I was producing.  So I consciously made the choice not to work on finding a partner, even though I knew it was the right thing to do.

7. Unrealistic Schedule, No Accountability

Before things started falling apart, I had an ambitious but reachable schedule for doing things.  If everything had gone exactly according to plan, I would have stayed on schedule and done great.  But there was a pernicious psychological feedback loop that hurt me.  When I slipped a teeny tiny bit on the schedule, I would get discouraged, and adjust the schedule backwards.  But when I missed that again, I started to feel defeated and ended up working less.  That made the schedule slip worse, which gave me less motivation to work, and this spiralled out of control to the point where I stopped.  In the middle of this cycle, I stopped announcing things on my blog because I was ashamed of how consistently I failed to meet any previous deadlines I set.  So I lost the reinforcement that comes from public accountability, which made it even easier to slip the schedule.

I truly believe that if I had been more realistic and flexible about my schedule (and about the scope of work, see #2), then I would have hit more deadlines, stayed accountable, and had more reserve willpower for when things got tough.  A track record of success would have gone a long way to encourage me.

8. I Was Shattered When My Plan Fell Apart

In 2009 things were going according to plan.  I had written a lot of code on my own, and I found what looked like a great partner that complemented my skills.  Things went pretty well for the first few months but long story short, that partnership died a slow uncertain death, to the point that it was very dead before I even realized it was dying.  Now it was six months later, there was unresolved uncertainty around ownership of the company and assets, and I a lot of the progress I thought I had made was now gone.  This was a blow that I couldn’t really handle and basically the death of the present version of GeekStack.  I didn’t admit it for a while because I wanted to be determined, I wanted to be committed, I wanted to be unstoppable.  But the thought of starting over and being so faaaaaaaar from launching was too much for me, so I just pretended like GeekStack was still alive.  I still read game design articles, I still collected scientist profiles, I still did little stuff, but from that point on I don’t think I wrote a single additional line of code on GeekStack.

Now What?

Writing this has been very cathartic so far, thanks for listening.  (What?  I’m the only one left?  Oh well, catharsis doesn’t need an audience.)  But getting things off my chest doesn’t matter if I don’t learn anything from it.

What are the good things about my experience?

• I have a well of about 6-9 months of determination
• I verified ideas in the market before spending too much time or money developing them
• I could develop code outside of a structured work environment and pick up new skills as needed
• I had interest and organizational skills beyond just coding
• I can expand a plan as needed to accomodate new information or ideas

What are the weaknesses I need to work on and be aware of in the future?

• I rarely was willing to scale down plans to something simpler or smaller
• I was very, very dependent on momentum and routine to stay motivated
• I would go into denial about problems I wasn’t sure how to fix
• I could get paralyzed now by problems that I didn’t need to solve until later
• By refusing to compromise schedule, I instead compromised completion and progress
• My short term goals were too big and too demanding

Action Items

I don’t know when or if I will return to developing GeekStack.  It has been nice to live without the anxiety of a big project hanging over my head.  I’ve spent more time with my wife and kids and been more focused at work.  I also really love the place I work now and I’m getting tons out of my time there.  I have been improving our money situation in slower, steadier ways than looking for a big startup outcome.

On the other hand, I love the idea of GeekStack and I still have a huge desire to improve the way society views creative, productive people.  I think the business idea is still sound and sustainable.  And the siren call of solving “the money problem” is still appealing.  So GeekStack is not dead dead, more like encased in carbonite.

If and when I start on GeekStack again, here are things I promise to do:

• Find a partner/cofounder to share the work with, or at least a Mastermind type group that can keep me accountable for progress
• Spend more time on task decomposition and prioritization, making sure that I do the most important things in order and that I have weekly-ish deliverables to keep visible progress and internal motivation high
• Get other people (customers, playtesters, etc) involved as soon and as frequently as possible, even if it’s in a more limited way than I hoped for.

I hope you’ve enjoyed this journey.  It was an expensive and roundabout way for me to learn lessons about my own psychology, but that’s a heck of a lot better than failing under a delusion.

—-

Got a similar experience or concern?  Need a shoulder to cry on?  Email me (my personal email) at peter @ pchristensen . com

Comments

About

GeekStack is making trading cards with the people, events, and achievements that our world is built on. Sign up now to help shape the first series, Computers.

Subscribe To The Blog

GeekStack on Twitter

Search The Site

Recently Written

• GeekStack Postmortem
• Balancing books + rubix cube + reciting pi = AWESOME!
• Nature by Numbers Video
• Science Has a Serious PR Problem
• Soccer Ball Generator
• Grand Challenges for Engineering
• TCGs are for fun, not pain
• On Science, Society, and Democracy
• The Myth of the Meritocracy
• Alternative Inspiration

Categories

• Biology
• Business
• Chemistry
• Education
• Events
• Game Design
• Geek Fun
• Geek News
• GeekStack News
• Kids
• Science
• Society
• Uncategorized

Archives

• August 2011
• March 2010
• February 2010
• January 2010
• December 2009
• November 2009
• October 2009
• September 2009
• August 2009
• July 2009
• June 2009
• April 2009
• February 2009
• January 2009
• December 2008
• November 2008
• October 2008

Links

• Peter Christensen