<M <Y
Y> M>

[Comments] (2) Loaded Dice, Round 2: At Pat's suggestion I did some more number-crunching and put the results in Loaded Dice. First, check out Standard deviation of ratings over time. Although the average rating is higher for new games than for old games, the standard deviation is always about 0.8. That is: the average rating for any 1980 game is some number, plus or minus 0.8; and the average rating for any 2010 game is a slightly higher number, plus or minus 0.8. The range of opinion is surprisingly limited.

Also take a look at the huge new section on Ownership and the Trade Market, full of graphs and tables taken from BGG's information about how many people own a game, how many people want to own it, and how many want to get rid of it. Including but not limited to:

[Comments] (2) : Whoever sets up the old-photography exhibits for the Metropolitan Museum is on fire. Earlier this year they put up "Our Future Is In The Air", an exhibit of photos taken in the 1910s, when cameras became cheap and portable enough for ordinary people to own them and carry them around photographing random crap. One frequently-lovestruck teenage boy put together a scrapbook of "Girls I Have Known" ('Titled, signed, and inscribed in ink on cover board: "GIRLS // I // HAVE // KNOWN [underlined] // D. ROCHFORD // DANGER KEEP OUT [underlined] // PrIVATE"; extensive inscriptions and attachments on 83 pages and on inside front and back boards.') Anti-child-labor crusaders surreptitiously photographed child laborers at work. And so on. Great exhibit.

In related news, I hit the Met again today and caught the next amazing photograph exhibit: "Night Vision: Photography After Dark". Since these photos weren't taken 100 years ago, they're not public domain, and a lot of the good ones are NO IMAGE AVAILABLE on the Met's website, but here are a couple good ones: "Broadway at Night", this odd untitled bit of meta-voyeurism.

"Night Vision" goes away next week but it's likely to be replaced by something else awesome. I'm not really into photography as an art medium, but whoever puts together these exhibitions is doing a really good job of grabbing my attention. Good job, anonymous curator.

Papers!: Recently I read some academic papers that were comprehensible and interesting, and since I'm not a professional academic, that's really all I need out of such things. Check 'em out.

First, "It was twenty years ago today", Paul Ginsparg's history of arXiv.org, which includes some history of the Web from the perspective of the community for which it was invented:

In early 1994, I happened to serve on a committee advising the APS about putting Physical Review Letters online. I suggested that a Web interface along the lines of the xxx.lanl.gov prototype might be a good way for the APS to disseminate its documents. A response came back from another committee member: “Installing and learning to use a WorldWideWeb browser is a complicated and difficult task — we can’t possibly expect this of the average physicist.” So the APS went with a different (and short-lived) platform. Meanwhile, the CERN website had partitioned its linked list of ‘all the web servers in the world’ into geographic regions, as if keeping such lists could still be a sensible methodology for navigating information.

Second, Scott Aaronson's tour-de-force, "Why Philosophers Should Care About Computational Complexity". Sample punch to the jaw:

Suppose we want to claim, for example, that a computation that plays chess is “equivalent” to some other computation that simulates a waterfall. Then our claim is only non-vacuous if it’s possible to exhibit the equivalence (i.e., give the reductions) within a model of computation that isn’t itself powerful enough to solve the chess or waterfall problems.

In other highbrow news, right now I'm reading a draft of Mike Amundsen's Building Hypermedia APIs with HTML5 and Node, and it's pretty awesome.

Indefinite Period of Kickstarter: There are still four unfinalized projects from my Month of Kickstarter project, so I won't be posting an analysis for a few days. I will say that it doesn't look good for three of those four projects, including one ("London in the 1980s") for which I'm still the only backer.

But in the month and a half since MoK I've still been checking that damn "recent projects" page every day and putting money towards projects. I haven't been doing two projects every day, or posting about it, but I did recently pass 100 projects backed (not counting the five that have been cancelled), so I thought it was time for an update.

One thing I did learn from MoK is that my linking to a Kickstarter project doesn't do a whole lot for its success, but I'm always happy to do what I can. So here's a few of the projects I've backed recently that I think you'd be interested in:

In other news, recently I was contacted by Larry Hastings, who wanted to start a Python podcast called "Radio Free Python". He searched the proverbial net to see if anyone had used that phrase before, and it turned out I had, in the examples for a presentation I gave at PyCon in 2003.

Displaying a cravenness some might consider unseemly, Larry offered to pay me a nominal fee for free-and-clear use of the name. Since "Radio Free Python" was a random phrase I came up with eight years ago to populate a sample HTML page, I rejected his money as syntactically invalid, and said he should back a Kickstarter project instead, telling me which one he backed. Well, he backed two: "Carnival - A deck and dice game." and the not-as-expensive-as-putting-a-hyperspectral-sensor-on-the-ISS "InTanj - A new software development methodology".

And the story has a happy ending, as the real-life Radio Free Python is up and running.

That's it for this edition of Indefinite Period of Kickstarter. As always, you can check out the Kickstarter projects I've backed, and the projects I'm interested in but probably won't back unless there's a chance I can push them over the edge.

[Comments] (5) robotfindspainting: Beth with her robotfindskitten painting Two exciting pieces of news from the long-dormant world of kittens and the robots who find them. First, Star Simpson has ported rfk to jQuery, proving that robotfindskitten remains relevant in today's world of Javascript frameworks.

Second, as you can see on the right, Beth Lerman painted a robotfindskitten still life! Here's a better picture on Beth's art site.

The still life has been in progress for a while. Beth and I collaborated to find NKIs that would work well in a still life:

(Punch bowl and punch not pictured.)

Beth presented me with the finished painting Monday night in a gala unveiling at my apartment, and it's now being classy in the living room. Speaking of classy, be sure to visit Beth's Etsy store.

[Comments] (6) Month Of Kickstarter After-Action Report: Crispy Duck Games Here you see the logo for Crispy Duck Games, one of the fictional studios in my novel Constellation Games (coming soon!). It's the first reward I've gotten out of July's Month of Kickstarter project. (Thanks, Brandon Eck; I didn't tell you anything except the name of the company, but you came up with something funny and appropriate.) Now that all the Kickstarter projects I backed have passed their deadlines, I'm going to take a graph-filled look at how they did, and take a few tiny stabs at guidelines for future projects that use Kickstarter or the Street Performer Protocol more generally.

Success rate

This is not a random sample of Kickstarter projects. I picked projects I actually wanted to back.

In all, I backed 52 projects for MoK. 31 of them succeeded, 19 failed, and two were canceled. My success rate was 60%, versus 40-45% for Kickstarter projects as a whole.

I think this shows good judgment on my part: I usually backed projects the day they started, and I avoided projects like MC Frontalot's music video that were obviously going to make their goal (or I backed them and didn't count them towards MoK.)

I don't have any information about the canceled projects, but I gathered basic information about the other 50: their funding goal, how much money they actually raised, how many backers they had, and how many updates their founders posted. This information became the graphs you're about to see.

I pledged at least $25 to each project, and more if necessary to get a cool reward, so my total expenditure was at least $725, and probably closer to $900.

This graph shows how much money each project raised. The line is the success line: projects on or above the line met their goals, and projects below the line failed.

Success rate graph

Let's zoom in on the projects that failed:

Failure graph

Check that out. Most failed projects raised pretty much nothing. But many raised thousands of dollars but didn't get any of it. If the cheese vat project had asked for $3k instead of $20k, they could have made their goal, and still been able to produce all their prizes. They wouldn't have gotten a new cheese vat, but $3k is better than nothing.

Why do projects fail?

I looked at the failed projects and came up with four classes of failure. These are my subjective opinions about what went wrong with the projects.

  1. About 30% of failed projects didn't hustle. The project creator put up a project on Kickstarter in the belief that hundreds of people would come by and back their project. Instead, they got me, the guy who looks at every single project despite the fact that Kickstarter really doesn't make it easy to do this. (eg. "50's Monster Movie Serial!")
  2. Of course, hustling is no guarantee of success. For about 30% of the projects, the project creator hustled, and got money, but not as much as they asked for. They should have used a different rewards system, or assuming they could have delivered the existing rewards with less money, they should have asked for less. (eg. the "Bursts of Light" anthology.)
  3. About 30% of failed projects clearly had both problems: they didn't hustle and they asked for way too much money. (eg. the Thousand Island dressing documentary).
  4. About 10% of projects hustled towards a reasonable goal, but didn't make it because the project or the rewards were too niche. I think the best example is the oboe chamber music recital, which offered oboe reeds or oboe lessons at the $25 level.

Not shown: a much larger population of failed projects that I didn't back, which may have failed for other reasons.

What is the hustle?

How can I judge projects based on a vague quantity like "hustle"? I'm using the number of updates posted to the Kickstarter project as a very weak indicator of hustle. Here's a graph of updates versus backers, for all projects.

Attempting to measure "hustle"

Kickstarter updates do not cause backers, if only because nobody but existing backers cares about your updates. But in my dataset, no project with zero updates ever got more than 50 backers. Updates and backers are both signs of this invisible third thing, "hustle". Updates are a good indicator that the project founder is hustling. Many update messages are exhortations to existing backers, asking them to propagate the project through their social networks.

It's quite possible to hustle and fail anyway. But if you don't find yourself writing some updates, it's a sign of a problem with your strategy.

Mean contribution

Here's a big difference between successful projects and failed projects. The mean mean contribution to a successful project is $76. That is, dividing the number of contributors by the money raised gives a certain number, and the mean of those numbers is $76. The median mean is $66. The standard deviation of the means is a huge $39, but I don't know if that has any statistical meaning.

(CAUTION: Sumana finds these graphs confusing, because the x-axis doesn't mean anything. It's just all the projects lined up next to each other. But I couldn't think of a better way to present this information.)

Mean contribution (successful projects)

The mean mean contribution to a failed project is $43 (the median mean contribution is $38, the standard deviation of the mean a somewhat smaller $23).

Mean contribution (failed projects)

I was pretty shocked about this. Even the numbers for failed projects greatly overshadow the $25 I usually kicked in (and still usually kick in). However, for a lot of board game projects, I contributed $40 or $50, because that was how much you had to contribute to get a copy of the game, and all those projects succeeded.

Takeaway lessons

  1. Start the really good rewards at around $50. (You can go a little lower if you're doing a book.)
  2. Try to get people who put a lot of money into a few Kickstarter projects, rather than people like me who spread it around.
  3. If you're not sure how much you can raise, try something in the $1k-$2k range.
  4. Hustle, dammit.

What's next

I'm a little loath to do this because it means a bunch more data entry, but I want to take a closer look at at least one of these projects, to figure out what "hustle" looks like in more detail. There's one MoK project that for me just defines "hustle", a project that reached a really high goal with a mean backer contribution of only $23. I figure that's a good one to investigate more closely. So stay tuned, or not.

[Comments] (3) Month Of Kickstarter After-Action Report 2: Details: I decided to take a detailed look at the tesselated cookie cutter project. I chose this one for a couple reasons. First, the project succeeded, so I'm not rubbing it in when I suggest things that could have been done better. Second, a lot of interesting things happened over the course of this project. Third, the product is cool.

I scraped the list of project backers to make this graph of backers over time:

Backers over time

The red diamonds single out days when the Keith Kritselis, the project owner, did an update.

This project was very popular among people who don't use Kickstarter. This pie chart shows how many other Kickstarter projects this project's backers (including me) have backed:

Other projects backed

Half of the people who backed this project signed up for Kickstarter just for this project, and never used Kickstarter again. These are the numbers as of today, so someone who joined Kickstarter because of this project but then went on to back another project would show up in the "1" slice of the pie, not the "None" slice.

Despite the broad appeal, the main reason this project succeeded was that something really anomalous happened on the last day. According to this update, what happened was Stephanie Nelson, who has a huge following as "The Coupon Mom", posted about the project on Facebook. This brought in a huge number of new Kickstarter users.

Here's another pie chart that looks at the Kickstarter experience of people who backed the cookie cutter project on the last day. 77% of those people signed up for Kickstarter just to back this project and never backed another project.

Other projects backed (last day only)

If you go to the bit.ly statistics page for the Kickstarter page's shortened link, you'll see the effect of Stephanie Nelson's Facebook post there as well. The vast majority of clicks on that shortened link happened on August 16.

This is not the first time the project had gotten a lot of publicity. Update #3 shows that in early August, the project got a write-up in the local newsweekly, the Austin Chronicle. But that only brought in ten new backers, probably because you'd have to type in a long URL.

Keith Kritselis hustled quite a bit on this project, both online and off; you can see this by looking at the content of the updates, and clever Kickstarter hacks like this:

Spread the word.

It's also clear that Kritselis's hustling worked: the project was funded. But it would have come up about $1500 short if the hustle hadn't grabbed the attention of one person with a huge preexisting audience. It's not smart to risk the success of your project on being able to find that person (unless you are that person).

Obvious equation and nonobvious corollaries

Your project will succeed if B*Cm>=G. B is the number of backers, Cm is the mean contribution, and G is the amount set as the goal. (British newspapers, feel free to use this section for meaningless filler.) You have control over all three factors. You set G directly, you increase B by hustling, and Cm depends on how you've set up the rewards system.

How big can B possibly get? Here's the number of backers for each successful Month of Kickstarter project:

Number of contributors for each successful project.

Obviously this varies widely based on your hustle and the appeal of your project, but realistically you're not going to get more than 350 backers, and 150 is more likely. The project all the way to the right on that graph is The Endangered Alphabets Project, which got a huge amount of press coverage from mainstream sources including the New York Times. That project has 533 backers. If your plan assumes you'll get 1000 backers, your plan is probably wrong.

Because your number of backers is going to be pretty low, you need to make each backer count. That means raising the mean contribution or lowering the goal.

Recall from last time that the mean contribution for a successful MoK project was $76, plus or minus $39. The best solution is to have a product that's worth $50-$100, and to make a $50-$100 tier that is the first one with a really cool reward. This, I think, is why board games do so well on Kickstarter.

If you don't have something that's worth $50, get that tier up as high as you can. Don't undersell your product! It will doom your project! It very nearly doomed the cookie cutter project.

Let me show you what I mean. Here's the data on the pledge levels and how popular they were among the 317 backers of this project:

The $5 is the first pledge level where you get something real in return: one set of the cookie cutters. At $10 you get two sets, and at $20 you get four sets. (The retail price of one set of cutters is given as $12.) These three levels account for 85% of the backers.

These prices are way too low. (Again, I say this knowing that the project was funded--your project might not be so lucky.) To fund this project using only the $5-$10-$20 funding levels, you'd need 550 backers. It needs to be more like $10-$20-$40. This is the part that I don't have data on, but I'm willing to bet that getting someone to spend $10 instead of $5 on your $12 cookie cutters is easy, compared to selling them on your cookie cutters in the first place.

BTW, if you sum up all the backers at the listed pledge levels, you get a total of $5800 raised—two hundred dollars short of the goal. So even with all the hustling and the big last-day influx and everything, the project succeeded because people pledged more money than was strictly necessary to get the corresponding rewards.

In short, the tesselated cookie cutter story is a story of bad incentive design overcome by hard work, generosity, luck, and network effects. But do some work up front and you won't have to rely so much on that stuff. Look at other projects in the same space and see how they succeeded or failed. Look at the tiers they set up, see how many people pledged at each level, see how much money they actually raised and where it came from. A cool video can get people wanting to back your project, but the reward tiers and the goal you set will determine how much money you see.


Unless otherwise noted, all content licensed by Leonard Richardson
under a Creative Commons License.