With Project Spaghetti‘s release, we’ve finally completed our one-year mission to make a game from scratch, showing everything that goes into making it from concept to completion.
As Eric always mentioned during the show, our goal wasn’t to make a tutorial, but to show everything in an attempt to lower the barrier some of us feel towards making a game and, we hope, get some people to go out and try it for themselves. That being said, a lot of work goes into taking a game from an idea on the back of a napkin to a finished product.
Though we’re continuing to brush up the game based on feedback or ideas to improve the experience, we’ve reached version 1.00, so it’s time to take a look back at the numbers in case they might be of some use.
First, we need to explain the difference between “real hours” and “task hours” for anyone who isn’t familiar with the way we’ve been approaching Agile and our task list over at Trello.
“Real hours” are the obvious units of measurement – the actual hours on a clock that went into the game. “Task hours” are the number of hours we expected things would take when we estimated them. As estimates never match reality, those two numbers are going to differ. With an Agile approach, it doesn’t matter how much they differ, as long as there’s data to calculate how much they differ and the ability to use that calculation for future planning.
Task hours were fairly easy to calculate because we had written every estimate on every task card within Trello. The tasks finished during the live game development show totaled 131.65 task hours, which comes out to about 2.4 task hours per episode (given 52 episodes, the Christmas Special, and the Music Special). The tasks finished outside of the show totaled 446.75 task hours, which translates to about 8.6 task hours per week. And, of course, adding them together yields: 578.4, or about 11.1 task hours per week.
Real hours are a bit trickier to calculate, because there’s no concrete record of how many were actually used, though we can give a rough estimate based on the number of hours per episode of the show and the number of hours we wrote down on the overtime task cards.
For episodes, we can calculate 2 hours x 52 weeks = 104 with a three hour Christmas special and two hour music special for a total of 109. Considering the number of task hours handled during episodes was 131.65, that means we tended to average about 1.2 task hours per real hour. In other words, we tended to overestimate how long things would take by about 12 minutes per hour. In fact, considering most episodes contained about 30 minutes of non-task related work (introductions, explanations, tangents, closings, animated shorts, etc.), the estimate could be further off, but whether we compensate for that or not isn’t as important as knowing how many task hours we can expect to take down per episode.
For overtime, counting up the number of hours we wrote down in Trello, we got 296.5 hours of overtime, which translates to about 5.7 real hours per week of overtime, versus the two real hours per episode. Given the number of “task hours” we did during that overtime: 446.75, we were handling about 1.5 task hours per hour during overtime. That’s about an overestimate of 30 minutes per hour.
Of course, our numbers might be a bit fishy, because they depend on us tracking time accurately, but the important thing isn’t so much the reality, but our trend when it comes to distorting reality. If we were to continue work on Project Spaghetti, we could, for example assume that 15 task hours worth of tasks would take about 10 real hours of overtime work to handle, or 12.5 real episode hours (6 episodes and change) to finish. That kind of information is useful for planning, and it’s what makes Agile and tracking burndown (the completion of tasks) a powerful concept.
Although we didn’t break the numbers down by task type (game design, graphics, animation, etc.) we know from experience that the graphics tasks were vastly overestimated (as we had no idea how long they would take and assumed the worst), while game design tasks tended to be on the ball or underestimated – especially when it came to introducing new systems, as opposed to modifying existing ones.
Combining the episodes and overtime, we ended up with a total of 578.4 task hours handled in 405.5 real hours, for a burndown of 1.4 task hours / hour.
How we could use these numbers to estimate future projects really depends on the ratio of episode hours to overtime hours. If we do another season like this one, with a ratio of 2.7 overtime hours for every episode hour, then we could probably assume it’s safe to take on the same number of task hours (with a cushion thanks to knowing what we’re doing the second time through).
Obviously things like struggling to figure out how to get GameMaker surfaces to work properly with scaled resolutions, how to save and encrypt game data, how to attach the game to Twitch IRC, etc. are behind us, but it’s probably best to keep that buffer just the same assuming the next project is more ambitious and carries its own, new unknowns. If, on the other hand, we were making a sequel to Project Spaghetti, we would probably far underestimate what we could do if we just followed the old numbers.
Now, let’s turn to the rest of the project. We’ve only considered tasks that have been written down and have been “done”, as in something has gone into the game, but there are obviously hours that go into planning and research in addition to the things that have a direct impact on the game. These numbers are even fuzzier to deal with – not many people keep track of how many hours they’ve thought about something, and our minds wander, so even if we tried, any number we could produce with a stopwatch wouldn’t be accurate, but at least in Eric’s case, it’s not too difficult a calculation to make thanks to a few assumptions: he does his thinking during the morning train commute (which is about 2 hours a day, of which about 1 hour is a “thinking hour” as opposed to a “battling businessmen elbows hour”). As evenings are usually two hours of elbow battling, let’s keep it simple and assume one hour per weekday for thinking along with an hour a day on weekends too for a total of 365 real hours over the year.
Research time is much more difficult to estimate, but assuming about three hours a week based on experience seems about right, for a total of 156 hours over the year.
The only two major “real hour” factors left are: the time taken to make the music, which hovers around 20 hours, and an extra hour a week to add to that fishy overtime hour total, for an extra 52 hours.
That brings our total up to 405.5 + 365 + 156 + 20 + 52 = 998.5 hours, just under a thousand hours in a year (almost 20 hours a week) to make the game that we released as version 1.00.
For anyone interested in how long it took to make the game, that should give you a good estimate, but let’s dive further down the rabbit hole and bring it out to what it took to make the show, handle production tasks, etc.
First, let’s first check out the time going into the live episodes. As the amount of time that goes into a live episode is already included in the game development time above, we’ll cut that out, but it’s safe to add an hour per episode (half an hour before and half an hour after) for setting up / cleaning up.
To edit an episode in Blender, up until hitting the render button, takes an average of 2 hours per episode (with some more intense edits like the Halloween episode or the Christmas episode offsetting easier edits).
Then, once the video is up on YouTube, it takes another three hours to annotate, listen and find the timecodes, and write the descriptions. Then an hour to update Trello/Pinterest/Twitter/Facebook, and, of course, our website. So that’s about 7 hours of overhead per episode per week, for a total of 378 hours over the year (52 episodes + Christmas and Music Specials).
Next comes the animated short that covers Eric’s break in the middle of the show. Composed in RPG Maker, it’s usually about two hours to write the story, last week’s summary, and the overall flow, then three hours to script & test it in RPGMaker, then an hour to record the video and render it in Blender, for a total of 6 hours per episode and a total of 324 hours (52 episodes + Christmas Special + Bonus Lost Episode).
In addition to the show itself, another 20 hours went into making the music, and 10 hours went into making the sprites and resources, bringing the total to 354 hours, though, to be fair, those hours started ticking down two months before our Season One debut.
In addition to preparing the animated episodes, the website, such as it is, has taken roughly 50 hours to get it where it is, with about 30 for the first setup and 20 for maintenance/site map updates/etc.
For Unplugged Episodes, the ones where Eric plays a game, they took about an hour each to record, an hour to edit, and an hour to post. There are a few still waiting in the wings, but we have forty of them, so that’s 120 hours.
The YouTube & Twitch setup and other related site setups took about 10 hours, banners and logos took about 20 hours, the infamous Line stamps took about 60 hours (about an hour and change per picture, plus time for posting, etc.), SNS activity was about an hour a week for 52 hours, resulting in an additional chunk of 142 hours.
Totaling everything up, we’ve got 378 + 354 + 50 + 120 + 142 = 1044 hours. Again, we’re around the thousand hour (20 hour / week) mark.
Hang with us; we’re finally reaching the end of our summations. Although some of the non-game tasks did take place in May and June before we kicked off the show on July 4th, 2014, we’re going to keep things easy and lump everything together: 998.5 + 1044 = 2042.5 hours.
As Eric really squeezed the most out of his time, we can assume that the most he can put into any project while holding down a full-time job is about 2042.5 hours.
But we’re not done milking the stats for potentially useful numbers yet. Ratios can be a good way to figure out where time is going, and for exploring where it should be going, so let’s start going into a few of those.
First, we can check the percent of time going into the game vs. the percent of time going into non-game tasks. For game tasks, 998.5 divided by 2042.5 yields 48.9%, leaving 51.1% for non-game tasks.
Breaking the game-design tasks down further, 22.4% of the total time spent over the year went into actual game creation tasks, 17.9% went into thinking/planning tasks, 7.6% went into research, 1% went into music. (or, of the time spent on the game alone, 45.8% went into game creation, 36.6% went into thinking/planning, 15.5% went into research, and 2% went into music)
As for the show production, 18.5% went into the live show, 17.3% went into the animated show, and 5.9% went into Unplugged episodes. (or, of the time spent on non-game tasks, 36.2% went into the live show, 33.9% went into the animated show, 11.5% went into Unplugged episodes)
Obviously, this is just the one-year benchmark and only covers the work that went into getting Project Spaghetti to Version 1.00, with no real publicity beyond a trailer and a special live episode, so more hours will eventually go into the project before it’s done, but a year and a first build seemed like excellent timing for doing some calculations and seeing where we should go for season two.
If you survived reading this down to here, we hope you got something out of these numbers too in terms of thinking about what might go into completing a game project. Obviously mileage is going to vary based on individual experience, tools being used, the game being made, the art style, etc. but we wanted to add at least one extra data point for people who are considering making a game.
We hope we’ve proven that, even if it takes time, it is doable, even working full-time, even with long commutes, even with families, especially when just considering the actual game design tasks, which, for us, clocked in at around 20 hours (19.2) a week.
Depending on your schedule and family obligations, there are several ways to divide 20 hours up: 4 hours per weekday; 3 hours per day if you decide to throw in weekends; or 2 hours per weekday and 5 hours per weekend day as some examples. The one thing we’d like to really stress if you are going to dig in for the long haul is having a task list, like the one we had over at Trello, so you can always break big goals down into more manageable chunks and keep yourself focused on one thing at a time, otherwise it can quickly become overwhelming. If you do break things down, you just have to take it a week at a time and you’ll get there!