Kickstarter

I rose late today after banging my head all night against some bizarre CUDA results, but those are stories for another day, in another forum. A yawn and shambolic shuffle later, I was surprised and delighted to find *The Finest Machine* 21% funded. We've officially left the <20% bottom-feeding level of "embarrassment" and are through the looking glass! Excitingly, most backings are now coming from people I don't know, though some friends new and old continue to toss in (backings from high school comrades like Shauna and Andrew were so unexpected as to move me to dig out the 1998 yearbook. Thanks so very very much.).

Why do I place a premium on 20%? Having looked through kickstarter's project browser for the past few days, it seems that 20% is about the minimum for a Publishing project to be featured on the front page (frequency of appearance appears dominated by absolute funding levels, especially due to the prominent "Most Funded" section) if ungraced (as we are) by "Staff Pick" status. So what?

My naive theory regarding Kickstarter economics is as follows:

(a) No one unaware of the project can fund the project. There might exist recipient-oblivious random funding sources, but I'm unaware of them, and frankly hope otherwise. Thus, funding must be a function of visibility. Let P be the number of distinct possible funders (i.e., number of people aware of the project).

(b) Some number P_f (0 <= P_f <= P) will fund the project at an average level of F_a. Our expected sum funding is thus F_s = P_f * F_a. All very clear.

  The first lesson we can take away from this, and something obvious before the campaign begins, is that your threshold funding level F_t had better be <= F_s. I set a threshold of $16,384 aka $2^14. Assuming an F_a of $32 aka $2^5 (F_a is actually currently around $44 for *The Finest Machine*), that requires P_f >= 512 (P_f is actually currently around 75). I figured I might get around P_f==128 backers; I ought then have aimed (via Reward levels) for a F_a of $128, especially as I assumed a funding population primarily composed of upper-middle-class engineers. Of course, at $128 I would no longer expect 128 backers. The uncomfortable result is that this project was unlikely to be funded without visibility beyond my own social circle. Ours is not to reason why; ours is but to do or die.

(c) Without detailed data spanning multiple projects, we can only conjecture about F_a and P_f. F_a lives in the realm of behavioral economics and There But For the Grace of God goes your Humble Natural Philosopher. P_f is rather more interesting. We noted it a subset of {P}; how is P generated? Each time a link L_n to the project is generated (whether by post or search result), some group P_Ln see it. P becomes {P ∪ P_Ln}, and P_nLn = {{P ∪ P_Ln} - P} new viewers might fund. Furthermore, P_nLn new viewers might add a new link. These new L_n will have their own unique P_nLn, of course, and furthermore increase the likelihood of search-derived links being returned for various keywords.

(d) We can express expectations of funding and link generation for each P_nLn (*not* each L_n, since visibility among various L_n is likely shared).

Now, people have been coming up to me the past few days, impressed by the project's progress and expressing expectations of successful funding. I've generally smiled and shook my head, pointing out that (1) we can probably expect the majority of L_n-resultant funding to occur soon after L_n is generated (darkly muttering something here about "low-λ Poisson distributions"), and (2) my own fairly large social networking reach means many L_n might be expected to yield negligible P_nLn, and (3) even if L_n-results are linear in time, the project has been linearly underfunded for all but the first two days. I've then noted, however, that the emergence of L_n from beyond my social network could drastically affect the outcomes. Since {P} grows via exclusive superpositioning of P_nLn, and F_s is determined solely by P_f and F_a, and allowing F_a to grow arbitrarily large reduces P_f, growth in P_f (dependent on P) is the only way to a win.

(e) it ought be obvious that hope of successful funding now rests entirely on L_n originating outside my social circle, or L_n targeted beyond my social circle (this latter made Donnie's post to Reddit particularly effective, funding-wise), either by exclusivity or simply a large readership.

I wish I had more data from successful and unsuccessful projects. My most pertinent question is: do successful projects typically see linear or exponential growth prior to meeting F_t? I expect that it's often the latter, due either to sudden increases in F_a (i.e. one or two megafunders) or sudden drastic increases in P_f (i.e. one or two links with major uptake).

Thus, crossing the 20% threshold (significantly increasing the chances of being front-paged) and doing so as quickly as possible has seemed of paramount importance.

Alternatively, I could peddle my still-young manflesh and attempt to seduce backers of extraordinary largesse, but in that case I'd likely just ask for cash directly. Don't let that stop you, though. The same principle applies to funding via the mining of bitcoins, equity speculation, a second job at Forever 21, or the firejousting my Samoan ancestors practiced aback rhinocerotes during the grim Days of Rage.

I'm sorry if this is all terribly obvious to everyone! Hack on! --nick