I believe the application of 250.122(B) as Norcal cites is a misapplication of the section and does not apply to a manufactured cable such as NM. The code section says grounding conductors in multi-conductor cable is to comply with Table 250.122, which makes #10 the minimum size grounding conductor for #6 NM @ 55A or #8 NM @ 40A.
Yeah, so what he said doesn't make sense and doesn't apply to the suggestion.
What someone suggested, which he was responding to, was to run 6-2 ahead of time, and then if the machine the OP purchases only needs a 30a plug, just change the breaker feeding the #6 from a 50a to a 30a. Norcal then stated there's a problem when over-sizing cable and using NM. Specifically, the #10 EGC in the NM cable would not comply with 250.122. But a #10 EGC with #8 or #6 IS the correct size because of the ampacity of those wire gauges! The NEC code he cited says that when increasing the size of ungrounded conductors, the EGC must also be proportionately increased in size. So Norcal- in THIS exact scenario, if the code you're citing applies, what size SHOULD the EGC be for a #6 cable? #8? That doesn't make sense!! A #8 EGC is for a 100a branch circuit!
What I think that code section has to do with is let's say I'm installing a 50a branch circuit and normally I would run #6 to feed it, but because of distance and possible voltage drop, I had to run #4. To comply with this code, I must also proportionately increase the EGC size because I increased the ungrounded conductor size!
In the suggestion for the OP's situation, cable size is not being increased due to voltage drop, its being re-purposed for a smaller size circuit!
EDIT: I agree with pattenp and oleguy in comments # 21 and 22. But now that pattenp pointed it out, I see what Norcal is saying. It still doesn't make it any more logical that a 30a circuit being fed by #6 requires a #8 ground! Plus, IMHO, that's a contradiction of NEC T250.122!