I think these two discussions have a few things in common.
The situation is different though, because it's one thing to say that we try not to use home brew as an explanation. In this case, we know for a fact that home brew is in play, and it's entirely different to try to explain things without home brew and to speculate on what the home brew contains. Since we 100% know he has a level in a prestige class and it's not unreasonable that the first level of dashing swordsman might advance bard spellcasting, I think the appropriate entry here is probably "Bard 14+, Dashing Swordsman 1+, Total Bard Spellcasting Levels 16+" since in theory either Bard 14/DS 2, or 15/1, or 16+/1 might work, depending on how Dashing Swordsman works.
Now, to the thornier debate (although really it seems to me that Ox is pretty much on an island here arguing for something no one else is buying):
I realize that this is getting into "that depends on what the definition of 'is' is" territory (and yah, I'm probably dating myself with that reference), but I think this is actually important. We take the word of Giant on actual stats/classes to be fact, but I don't think even "I wasn't thinking of Tarquin as a swordsage" is actually any evidence at all.
What the flowchart that Pelee posted means is that the Giant doesn't care, but the important thing to remember here (and the source of a lot of frustrated posts when people forget this) is that the comic, as conceived, doesn't actually operate under anything like RAW of 3.5. We therefore aren't in the business of figuing out what the Giant most likely has in mind, we are in the business of determining--as best as we are able through necessarily somewhat vague and occasionally contradictory rules--what set of character traits best explain what we see on panel.
More precisely, we are in the business of determining what stats can be reasonably inferred from what we see on panel, no more, no less.
In general, as Rich has gotten less interested in rules over the last 20 years (?!!), this has gotten a lot harder to do. The panels get less detailed about mechanics, and Rich no longer tells us what's going on mechanically. For this reason, I am personally all in favor of coming up with some sort of system where we link to past debates or include gray or spoilered "most likely but not meeting thread evidentiary standards" or some other system that someone who cares more and/or is smarter than me comes up with.
That said, there's a very good reason we didn't set out originally to determine most likely stats (as opposed to stats that can be clearly shown), and it's because, as this debate and others have demonstrated over the years "most likely" is grounds for incredible amounts of unresolvable differences, so the less arbitrary and strict we make the rules, the more confusion and disagreement we get on the ground.