Thor, Luke,
I was wondering why you guys went with a random chart rather than selecting spells for the arcane casters in Torchbearer? It seems like an odd fit given the intentional choices made throughout the rest of the process.
Thanks
Note that finding a specific spell has no guarantee whatsoever in the game. You either have to hunt up the scroll, or find someone in town to teach it to you. So this is the same; magic is scattered enough that you get what you get. (Also, don’t forget that magician is hard mode.)
It’s also a throwback to OD&D [citation needed]
I totally get that it is a throwback, but I’m wondering if it somehow supports other design goals that I’m not seeing. And I can’t help but think there has to be a better way to support the goal.
IANT, but I’d think it:
[ul]
[li]Speeds character creation, by avoiding analysis paralysis, which in turn supports the high lethality of the system.
[/li][li]Similarly, enhances replayability and verisimilitude by provide high variability (i.e., no “Yeah, this is Bob… he’s a LOT like Rob… who was a LOT like Nob.” It’s Torchbearer not Paranoia.).
[/li][li]Forces creativity, by providing constraints and unusual combinations and avoiding “Ogg Mages”.
[/li][li]Similarly, it mitigates the inherently unbalanced efficacy from spell to spell (i.e., one isn’t incentivized to pick the same set of ‘best’ spells).
[/li][li]Supports a tone that magic spells are hard to find and masters are fickle.
[/li][li]Provides immediate incentive to quest for magic you do want.
[/li][/ul]
…and that’ just off the top of my head, without having finished reading the rules! 
Another great benefit: it encourages you to use magic creatively! We had a situation just like that, where a player used Word of Binding (or whatever it is) on a trapdoor; the spell had previously been judged useless.
Exactly, and a great use of wizard lock.