I'm facing the same dilemma with a hardware project of mine (http://tooner-test.moshozen.com). In the past month alone, I've been stuck on several things (public name, dealing with constant ID creep, finding a mill that can resaw, among others). Though they're contradictory, it seems that half of the roadblocks get solved through putting them off (and usually thinking of a better solution, or a workaround), and half get solved by #JFDI.

'Shipping' means something much different for hardware projects of course, but nonetheless I think the advice to ship on day one is really foundational. I've always been fond of the idea that 'if something hurts, you need to do it more often'. Make the game about iterating and not shipping.