Lead them hand in hand,
Show them every small detail.
They won't build, just bail
On a plane to Bali for a conference (why we're flying East I'll never know), so it looks like I can actually put some time into writing (this, as opposed to the other papers that I really should be working on instead).
OpenHand remains a stick in my side. On a good note, I'm close to refining and releasing a whole slew of design variations, some of which should be much more obtainable in terms of cost and fabrication, so hopefully that helps pull in the users still hesitant to fully commit. That said, as we're iterating on these designs more and more, some of the insurmountable challenges of printed parts and off-the-shelf, hobby-level actuation schemes are becoming more apparent. At this point, I think comparing our hands to commercial hands (like the BarrettHand or Robotiq series of hands) is liking comparing dirt-bike to a Harley. You may be willing to pay top dollar for the quality of the latter, but you can't push it to its limits like the former. That said, every once in a while you'll find yourself walking your dirt-bike home instead of riding it.
That then begs the question: why get something that needs continuous tuning and adjustment? Lower cost can't be the entire story. I think it boils down to accessibility and flexibility, and this becomes especially apparent in the DIY communities' call for open development. An open-source base should set the stage for vibrant development, but only with active participation, so how do we make that happen? I figure a more thorough examination of the big players in open-source may help:
Admittedly, I'm not the best qualified person to discuss the merits of Ubuntu, especially since I don't quite get the (recent?) backlash against it from supporters of the other Linux distributions. However, I'd say it's one of the best examples of an open-source project matching (and in some ways surpassing) the success of the commercial top dog, Windows. It's almost common-place now for me to see an Ubuntu installation option next to those for Windows and Mac OS. Then again, the closer it's come to matching Windows in terms of usability (assuming that was the goal all along), the more of its past advantages it seems to have lost. I no longer can run it on my netbook due to performance woes, and it seems to get more bloated with each new release. It's almost as if now it's just an alternative product more than anything else. Does open-source success result in convergence? Or did it just happen in this case because the first-to-market pretty much dictated what the users want?
A similar scenario pops up every time we discuss our underactuated, flexure-based hands with more traditional roboticists. They'd argue that they need proper joint models and the option of independent joint control, and we'd tell them that's not the point. When a design is open and begins as model A, but the bulk of its potential users want model B, does A eventually become B? Or does it have a shot at becoming something different?
Sometimes I'm not sure I even consider Arduino to be open-source, as the standard they've established is so ingrained at this point that they're the only ones who can introduce future iterations (there's got to be some weird economic term for this). They're like Lego in that they sell building blocks, except they specialize in that one particular building block that everyone needs. I think it's interesting that their one block does several jobs only sort of well, and yet that's more than enough.
Their commitment to accessibility may be unmatched, and their community has detailed everything from the simplest of beginner projects to the weirdest edge cases. To steal an overused term from the startup world, Arduino's real business is in minimizing the friction between start to finish for its users. Plenty of clones of the Arduino electronics exist, but that's not really what Arduino is selling nowadays. Minimizing friction can't always be achieved by product refinement alone. Strange that hardware is now doing what software startups have been abusing for years: tweaking the user interface to long-standing technologies.
Fucking Ikea. Corporate behemoth using 1% of the world's wood (hehe, they got wood). The hallmark of their open-source success is an independent website that they actually tried to shut down. All that, and they've got the best open-source platform in the world (in my opinion). Extensive build documentation on every page of their website, sets of compatible parts that can be mixed and matched in ways the company can't even predict. Multiple products share common components and can accept 3rd-party pieces, and users can put most of these products together with minimal experience and the most basic of tools.
It's no coincidence then, that we often compared OpenHand to Ikea when discussing our approach and goals. It's not enough to just put out the plans for a functional hand. Users don't just need their hands held. They need the necessary parts gift-wrapped too. And yet, I think the big key to Ikea's success is restraint. Despite their extensive library of components, it's not like you can generate any furniture out of the blue from their design set. Various components and pieces have been retired over the years, and while I'm sure they're mostly profit-driven decisions, the loss of flexibility results in less confusion (the paradox of choice) and increased accessibility. Users don't want a table that can be adjusted to any height. They want the table that perfectly fits their needs. There may be more than one way to skin a cat, but do anyone really want to skin cats at all?
PS - writing this while re-watching Avatar (James Cameron, not M. Night). OMG the plot holes. OMG.
PPS - started watching Gotham while finishing this up. Feeling kind of meh about it. To be expected.