Once you know that this started life as a sequel to Diddy Kong Racing, it's quite striking how many surface-level similarities there are between this and one-third of DKR. The controls for piloting have been fairly faithfully translated to GBA, with only a little bit missing; I definitely miss being able to quickly steer by holding "R", but the fact that they were able to implement stuff like the loop-de-loops and aileron rolls into a sprite-based isometric game is pretty impressive. There's also little stuff like the appearance of the dash rings, or the inexplicable Smoky the Dragon cameo, or the Jiggy Challenges being parallels to the Silver Coin Challenges, or the fact that one of the characters (Grunty) is incomparably worse than the rest of the cast as Krunch was in DKR.

Banjo-Pilot had a pretty rough dev cycle, only in part due to the Microsoft buy-out. It's fairly well-documented that there were at least four different iterations to this game before they settled on what was released. To be honest, I think the team spun their tires on this game because they genuinely couldn't find a good way to make the project work. Diddy Kong Racing for GBA is a super ambitious idea, one that frankly wasn't possible for the hardware. The closest attempt to realizing this - the 2004 build done in voxels - apparently ran so slowly as to be unplayable, which is why the team reverted back to their 2003 build and cleaned that up for release. I suspect they just needed to get any version of the game out by 2005 to get their return on investment.

Unfortunately, it's not hard to see why the team didn't originally commit to the 2003 build. It's not bad, but there's just so little to Banjo-Pilot to make it stick out. 3D was what made flying interesting in Diddy Kong Racing, and the Mode 7 illusion of 3D simply isn't able to replicate the feeling. There are so few obstacles to contend with, and the tracks are so barebones, that races are just over and done with so quickly. I don't think any track lasts longer than 2 minutes? And the difficulty exists entirely within the character you choose, more than anything offered by the track or the AI. There's no "rival" system in place, the way it exists in Mario Kart, so it's ridiculously easy to run away with Grand Prixs and the like.

Relatedly, nothing whatsoever is done to balance Bottles versus Grunty in the Jiggy Challenges. So your reward for struggling through all the Bottles Challenges against the most broken character in the game is a piss-easy rehash against the worst character in the game. I generally find that the Reverse tracks add little to the overall experience, largely because the tracks themselves have little mechanical identity such that a Reversed run would change them up. I think Freezing Furnace is the only track to express a clear mechanical idea (splitting everything across those three-to-five channels along the straightaways), with everything else simply being about the shape of the track.

It is sorta interesting that this is the game where Jolly Roger first started to get a push. Jamjars would've felt like a more logical inclusion, since Humba Wumba's there already herself. But I'm not inclined to complain or anything, I'm all for the more off-beat character inclusions. Even if they don't exactly make sense; why does Bottles control lightning in his versus matches? Or Banjo launch his WISHYWASHY incarnation at people?

Mario Kart: Super Circuit feels like a natural point of comparison, in much the same way Diddy Kong Racing invites natural comparisons to Mario Kart 64. Super Circuit is an arbitrarily mean game, with sloppy controls and a strict definition of what it considers mastery. Banjo-Pilot is simply a title without teeth. Maybe those Staff Ghosts require getting good to clear, but I wasn't really interested in trying to grind those out. Stop 'n' Swop is dickish here, but that's sort of the joke as well. Everything else just... kinda exists. I imagine this would've made for a much more enjoyable portable experience than Super Circuit back in the day, by virtue of it being easier to master, but it's so nothing that it doesn't much feel like an improvement. A decent experiment, but kind of a failure, sad to say.