Prototyping only means something if you’re actually willing to change your mind when real use tells you something the spec sheet didn’t. Here’s what’s genuinely changed on the sleep system since the first versions, and why.
The quilt’s baffle structure
The first versions used long baffles running the length of the quilt. In use, down migrated inside those long channels over time, settling unevenly and creating cold spots exactly where you don’t want them, in the middle of the night, when there’s nothing convenient to do about it. We moved to a grid-baffle structure instead, smaller, more numerous chambers that keep the down far more evenly distributed.
It adds a small amount of extra stitching and a small amount of weight compared to long baffles. We think that’s a genuinely worthwhile trade for not waking up to a cold patch that used to be full of down before it drifted.
The quilt’s footbox
The original footbox was permanently closed, sealed for maximum warmth and draft protection. In practice, that made it too restrictive for a lot of real use, anyone who runs warm, or wants to vent on a milder night, or just wants to stick a foot out, had no way to do that without unzipping the whole system. We moved to a more open, flexible footbox design. It’s a small step down in pure draft-proofing compared to a fully sealed footbox on the coldest possible night. It’s noticeably more usable across the actual range of nights people use the quilt for, which matters more in practice than performance on the rare extreme.
The pad’s weight
This one’s less flattering to admit, and we’re saying it anyway. An earlier pad prototype came in around 610g. The current build is closer to 666g. That’s not a typo and it’s not a rounding error, it crept up as the pad went through refinement, most likely changes to the valve, fabric or seam construction along the way. We’re not going to pretend the number stayed flat when it didn’t. It’s still provisional, and it may move again before this is finished.
Why we’re saying any of this out loud
None of these were mistakes exactly. They’re the normal, slightly unglamorous cost of testing something against real use instead of just trusting a spec sheet from the first prototype. A design that never changes after the first build either got everything right immediately, which is rare, or nobody was actually testing it hard enough to find the problems. We’d rather be the second kind of honest than the first kind of convenient.
Over and Out