What’s Getting Backed on Kickstarter: Technology Edition

I’m a huge Kickstarter fanboy. The creativity that they’ve unleashed is mind-boggling and embodies exactly what the Internet is capable of.

The projects I’ve backed tend to skew tech-heavy and I recently wondered if there were any identifiable trends as to what sort of tech projects get backed. I hear all the time about companies that started with a Kickstarter campaign; are there any patterns?

Kickstarter is really open about what projects get backed (note that it’s not clear how accurate the project counts are; they seem to appear/disappear/dramatically change based on when you access the site), so I went out and scraped some data (code here).

I managed to get data on 1700 successful tech projects, which raised a cumulative total of $212,472,913. That’s an average of $124,984 per project-but this is no even distribution. The winners (like the Pono music player, Reading Rainbow or Zano drone) raise millions while Kymira smart sports apparel is limping in at $8,000 or so (but still successful; kudos). The median raised is $45,992.

I wrote some code to try and cluster these projects and found a few categories that people like to back:

Physical Computing: Arduino clones and shields; Raspberry Pi accessories galore. Examples include Microview, RFDuino and the Touch Board.

3D Printers: Every type you can imagine, including the Micro, the Form1 and the 3Doodler.

Home Automation: Smart plugs, dimmers, remotes – Kickstarters want a connected home. Sample projects are Ube Wifi dimmer, the NEEO remote and  the Ninja Sphere controller.

Lighting: Make it glow-whether lights attached to your stereo, fancy bike lights or an enhancement to your GoPro. Examples are the Notti Smart Light, Lume Cube flashbulbs or Playbulb candles.

Phone Accessories: Anything that can pimp your phone. Check out the Jorno foldable keyboard, Thermodo thermometer or Chipolo item finder.

Solar Powered Gizmos: Kickstarter backers seem to really want to take their electronics outside. Witness the WakaWaka Base, SPOR and Solarpod Pyxis chargers/lights.

Here’s how many projects fall into each category:

That’s a high of 433 for Home Automation versus 181 3D printers.

(Hate that there are no numbers in these graphs? I do, but can’t figure out how to add labels. All the raw data is here.)

The total funds raised varies dramatically across the categories ($M):

Surprisingly, Physical Computing has clocked almost $50M ($46.4M), closely followed by 3D Printers at $41.4M. Solar Power is half this at $21.5M.

There’s a similar difference in the the amount raised per category-both the average and median ($K):

On a per-project basis, 3D Printers have captured peoples wallets (most likely have a much higher per unit cost than other projects) and clock in with a $229.0K/$79.9K average/median raise. This median is almost as high as the average for solar power projects: $89.4K/$42.3K average/median.

A couple of closing thoughts:

  • I was amazed at how much money has gone towards backing Physical Computing. I imagine that most of these devices were bought by geeks to make geeky devices, so I’m guessing that we’re only at the starting of a big revolution in Internet-connected devices
  • There was no major cluster for robotics or drones. This surprised me as I would have guessed more based on the buzz in the press. Big difference between what is bought vs. what is talked about
  • Some of the major success stories (like Pono or Reading Rainbow) don’t fall into an of these categories. I don’t yet know how to interpret these “one hit wonders” but its interesting to think about why they succeeded as a product but didn’t launch a category

Don’t Be Fooled By Black Swans

Last night Wen, Rich and I went and listened to Nassim Nicholas Taleb be interviewed at the powerHouse Arena. I’ve been a big fan of his books for years, but this was the first time I’d seen him speak. The interview started on a couple of false notes (he spent a few minutes telling us that they’d all just been out for drinks; the interviewer apologized for her French-accented English), but he had a couple of quotable points:

  • The difference between a fool and a saint is timing
  • If a problem is too hard to compute, the outcome is essentially random
  • Black swans are not black swans for everyone: only for ‘suckers’. To be crass, the 9/11 terrorist attacks were a black swan for Americans; for the terrorists were exactly what they were expecting
  • Debt levels map one-to-one with forecasting overconfidence
  • If I told you that you have a 3.4% chance of losing everything on a trade, you probably wouldn’t take it. If I told you that a catastrophic failure only occurs every 30 years, you would
  • Religion is not about beliefs, it’s about creating heuristics for people who otherwise couldn’t think them up themselves
  • The best science is done by independents (Einstein, Darwin), not by people associated with institutions – those people try to please the tenure committee. There probably isn’t a perfect institution for creating better science, but abolishing tenure is likely a good start. (This feels very akin to how innovation in business occurs)
  • ‘Forecast’ is ‘prophesize’ in Arabic – but how would you feel about next year’s business ‘prophecy’?

Basically, everything he said could boil down to the following:

  • Almost everything that’s interesting in the world is nonlinear
  • And no one really understands how nonlinear dynamics work
  • So if anyone tells you they do, don’t believe them
  • Instead, always compute the likelihood that something will happen…
  • …and make sure that you’re never the ‘sucker’ based on those probabilities

He closed with an interesting comment that he wants to move from a world of true/false to sucker/non-sucker. An interesting thought; if you get a chance to see him speak, do so.

Transient City

If you follow this blog, you’ll know that we launched UncoverYourCity to help people better understand New York.  Here’s another example of interesting things we can learn about the city by playing with the data.

Let’s say you wanted to see the relationship between people renting and their household income.   A few things emerge here: if you’re, on average, poor, you rent.  If you’re rich, you buy.  The poorest Districts of New York are in the southern Bronx and a shocking 92% of people rent; just up the Harlem River, 94% of people in Morris Heights rent.  The average family income here is around $17-22K per year.

The converse is also in the south: in southern Staten Island, 85% of people own and the household income hovers near $83K.  Here’s a comparison of all the districts side by side.

However, if you look at the plot of household income vs. rental rates you’ll notice a bunch of outliers in the upper right corner:

These are a couple of districts towards the tip of Manhattan (districts 1, 2, and 4-8) plus district 6 in Brooklyn (Cobble Hill, Carroll Gardens, Park Slope).

What’s going on here?  Well, if you asked me to pick the areas in New York where people who are passing through (i.e., getting experience before they move to a different city or start a family in the suburbs) live, I would pick these neighborhoods.  These are places where younger professionals move to get experience yet maintain a high quality of life (yes, there are lots of people there who are neither young nor professional, but we’re talking ‘on average’ here) and this may be why we see this litter cluster of outliers.

What’s interesting is what’s not in the set of outliers: Williamsburg (Brooklyn District 1) and the Lower East Side (Manhattan District 3).  My bet is that these neighborhoods are both subsumed by the averages of their districts and their more bohemian younger residents may have a lower average family income.

This data doesn’t tell the whole story, but it does help us understand the city a bit better.  More fun examples in future posts.

Healthy Things

I’m always amazed by what correlates with education.  The latest statistic I’ve learned about is regular exercise.  It appears that the less education you have, the more likely you are just to laze around:

Regular Leisure-Time Physical Activity

This data comes from the CDC and is technically the percentage of adults aged >25 years who reported regular leisure-time physical activity.  Two things immediately stand out here:

  1. If you’re poorly educated, you’re not exercising (and therefore, I’m going to go out on a limb and suggest more likely to be obsese)
  2. The trend is incredibly disturbing: only the well educated are engaging in more physical activity now vs. 10 years ago

I can’t help but wonder if this is going to become a self-reinforcing process.  Let’s speculate: if you’re poorly educate, you’re probably poorer.  Therefore you might have to work two jobs/more hours and have less time for leisure time and be more tired.  You probably also have less disposable income and therefore eat more fast food.  You’re getting squeezed by the current economy (and the long term trend that Chinese/Indian/Mexican workers are always going to be cheaper than you…).

If you’re highly educated, you can afford to eat better and have more job opportunities to keep your salary up while you are enough of a commodity to maintain a work-life balance.  What’s more, is that there are a whole host of services for you to keep you fit: gyms are an obvious one, but more interesting are the emerging class of high-end devices like the the LifeScan diabetes iPhone application and tracking tools like the FitBit.

This will be one long-term trend to monitor; hopefully the CDC will get us data more than once every 10 years.

Fun With Stats

I know it’s almost impossible to have fun with stats (I’ve sat through many a boring stats course over the years), but a recent event reminded me of how misleading some statistics and analyses are – most notably anything involving  time series and percentages.  There’s a small library of books on how to lie with statistics, but the recent announcement of the Palm Pre provides a great example.

For background, Palm’s a company that was synonymous with mobile computing, until Microsoft got into the game with Windows Mobile and then two companies called RIM (Blackberry) and Apple came along.  For the past few years, Palm has drifted and has been consistently losing market share.  However, on January they announced the Pre and this propelled their stock price into the stratosphere:

Palm Ticker - Jan 6-9, 2009Note that I’ve compared them with Sprint (whose share price went up as they’re the exclusive carrier for the Pre in the short term) so that you can see the percentage change (all the data comes from Google Finance).  This graph is a dataset with a few hundred datapoints and a naive analysis would suggest that Palm was a go-nowhere stock until it announced the Pre.

However, the truth is a little more nuanced than that.  In fact, Palm was up over 100% in the first part of December and has been on a tear since January 5:

Palm Ticker - Dec 5 - Jan 9As an aside, it’s interesting to note the run-up in the stock before December 22nd, when Elevation Partners invested $100M in the company.  People who work at public companies go to jail for trading based on this sort of knowledge, yet somehow the market was able to guess that this was going to happen.  On Friday the 19th, analysts couldn’t figure out why the stock price was going up, and then on Monday the 22nd they got the investment.   Hmmm….

So, now you’re thinking, Palm must have been great to their shareholders – I mean, they’re up over 200% since early December – that’s awesome, right?  Except that 2008 was such a bad year for them that they were down almost 80% for the year by the time December started, and now they’re almost back to scratch:

Palm Ticker - Jan 4 '08 - Jan 9 '09One of my favourite lessons regarding percentages is fully embodied in the graph above: if you’re down 80%, that means that you need a 400% increase to get back to where you started.  This is one reason why the current financial crisis is going to be really painful for people who bought at the peak.  You need a lot of “20% rallies” (about 7) to get you back to where you started if you are down 60-80%.

Speaking, of which, let’s look at how Palm has done over the years.  They IPO’d back on March 2nd, 2000.  They pretty much were the peak of the Internet bubble (another aside: through the magic of finance and arbitrage the publicly floated part of Palm was briefly worth more than its entire parent – which still owned 95% of the company):

Palm Historical TickerThey’re down a whopping 98.9% since then, so if you bought and held, you’re almost certainly never going to get back to where you started (unless they announce a heckuva lot more products like the Pre).

If we go back a little further, we can see one of my favourite facts about percentages in Sprint’s share price.  For background, Sprint was a darling of the New Economy and gained almost 100% between 1999.  It then declined and is down almost 99% over the past 10 years:

Sprint TickerTwo things I love here:

  • If you’re up 100% and then you by twice what you gained (i.e., a $1 stock goes to $2 to $0.01), you’re not down, 200%, you’re down 99%
  • People think there’s very little difference between a stock that’s down say 99% vs. 99.5%, but in fact the change is massive.  The stock has to drop by 50% from when it was already down 99% to get to 99.5%.  ($100 -> $1 -> $0.50).  This is trivial if you’re already down 99%; soul-crushing if you bought in when it was down 99%.

So there you go, some Sunday evening stats.  Hopefully it wasn’t too painful and maybe even insightful…