Hey, everyone. So, yeah, it’s been a while. Again. Like almost exactly a year. My last post was about the Kickstarter GoCube campaign, and here I am again talking about the same thing. See, the campaign funded, they produced the cube, and I’ve got one on my desk. The technology is pretty incredible; they crammed a lot of stuff into a cube that still manages to perform pretty well.

I don’t do unboxings all that often, but I wanted to support the campaign and get back into the swing of things on this blog and my YouTube channel. So, why not? Here’s the unboxing-slash-review, with all sorts of comments following below the fold:

Pretty neat huh? For those more into reading, here’s the good, the bad, and the ugly:

The Good (Amazing Tech)

Let me start by heaping praise. I know I highlight some negatives below, but it’s only because the tech is so compelling and well-executed that a few flaws emerge by comparison.

The cube physically performs quite well. It corners cuts at 45º+, has a nice clicky feel from the magnets, and is very controllable. Some of my double flicks did overshoot a bit (which can mess with timing), but it took only a couple solves for me to adjust. Considering how much tech is built into this thing — accelerometers(s), sensors, LEDs, bluetooth radio, etc. — it’s a masterpiece of a speedcube.

The tech itself is flawless. The app connects to the cube seamlessly and consistently; the app reflects rotations and turns with only the smallest of delays; and the battery life seems really impressive. The digital representation of the cube on the app is excellent and realistic. The accelerometer is the biggest improvement over past attempts by others to make connected cubes, and it makes all the difference. There’s virtually no discernible difference between the cube as it appears on the screen and the cube (including every cubie, orientation, etc.) as it exists physically in space. What a huge accomplishment.

While I heap some some constructive criticism on the software’s interpretation of some of the data coming from the cube, I feel very strongly that it’s all on the app side — and can be improved through updates.

This cube is damn impressive technologically. The designers and engineers did an incredible job.

The Bad (Some Software Deficiencies)

The software is really slick and clean. I haven’t dug through every part of the app yet, but (some small UI issues notwithstanding) it’s sharp and well-executed.

I’m particularly impressed that the app recognizes progression of a solve through the CFOP stages of cross, F2L, OLL, and PLL. Breaking down solves into those chunks — including separate treatment of each F2L pair — is very, very slick and unexpected.

That said, as I highlight in the video (around 9:00), there’s room for improvement in the software. There are at least four major improvements I’d like to see:

(1) Wide Turns, M slices

The app does not recognize either wide turns nor M (or S) slices. While there is equivalence between, say a L and a Rw in the sense that the middle slice and right layer stay together while the left moves relative to it, there is a major difference in solving technique and orientation. The same is true of M slices. There is equivalence between the middle layer moving against the stationery outer layers, and the two synchronized outer layers moving against a stationery middle layer. But, again, there’s a significant difference in technique and orientation. With an accelerometer in the core — I assume it’s in the core? — orientation changes (vel non) should allow the app to distinguish between these equivalents. It would make a critical difference in properly notating turns. This was my biggest complaint with other connected cubes, and one I thought Go Cube would triumphantly solve; yet, even with an accelerometer, it persists.

(2) Cube Rotations

The same is true of cube rotations. For example, the app does not recognize my V Perm — z D' R2' D (R2 U R') D' (R U') (R U R') D (R U') z' — as being preceded by a z rotation. Instead, it records it as D' D' B D D F D' B' D F' D F D' B D F' F', which I don’t think translates even if you adjust for a z rotation. Something’s quirky here.

Even more generally, the notation seems to be off during large parts of the solves by not recognizing rotations and then adjusting around them. I’m still working through the culprit, as it doesn’t seem as simple as the app being stuck with white as U and green as F. It’s a more dynamic/complex bug.

(3) Missing Basic Timer

One pure UI/UX crticism: As many bells/whistles as the app has, it’s missing what I would consider the most basic and necessary component: a simple timer section that gives scrambles, records solves, allows for sessions (including session stats), etc. See Five Timer as the gold standard. I should be able to do an Ao5 session, then get a simple breakdown of each scramble, turn-by-turn solve notation, and solve time (including, with nifty allocations of time across the CFOP stages). The sessions should be save-able, name-able, and share-able.

(4) PLL/OLL Recognition

Given the surprising CFOP stage recognition (seriously cool!), the app should inlcude a reference database of OLL and PLL (and even F2L) cases. What I mean is that the solve notations should recognize that the OLL case was, say, #34 and the PLL was an R Perm. That would be a killer (and I think not terribly difficult) addition.

The Ugly (Some Physical Quirks)

Just a few quick comments on the physical build of the cube. As mentioned in the “Good,” the performance of the cube is very impressive. I would have judged it against just being adequate (with all the tech), but it’s well beyond adequate. But I have two comments on the physical construction:

First, the beveled edges are really strange. Although I got used to them, they don’t have a great hand feel, and I think that contributes to some mis-timed, out-of-sync turns. I don’t see any reason it had to be shaped that way to fit the tech, and I would like to see a future version that’s squared off a bit more.

Second, and this is super picky, the colors are a bit dingy/muted. That may be just in the eye of the beholder, since I have a weird color scheme — fluoros on the sides, gray up, and black down — but I do think brighter colors would be more attractive and help with recognition. Given that the cube can be illuminated — the evasive grail, I’ve hacked my way in search of — it would have been really cool if the colors were a bit more translucent. In the dark, the red and orange are indistinguishable, while blue looks like black. Hey, Go Cube, any chance you’ll offer replacement tiles for the faces in different colors/shades/transparency?

Yes, I’d Recommend it!

Despite some software-side bugs and a couple physical quirks, this thing is damn impressive. I would highly recommend it as-is. And I suspect there’ll be big improvements over time that will close the gap between expectations and current execution. I’m such a big supporter of well-executed crowd-sourced and crowd-funded innovation, and I think this is an example par excellence. Go get yourself a GoCube!