I think the law of averages would take care of that. Bandwidth is pretty cheap and the chances are that even if you are constrained by bandwidth, as might be the case with a smaller site on an "xGB/day" hosting plan, then it's more likely to be the case there won't be too many GB of content to spider in the first place. There are always exceptions though, and where there is a real problem there are still going to be workarounds, e.g. explicit opt out clauses for spiders like IA's or, if all else fails, denying access based on User-Agent strings.
It does clearly depend on what effect this might have on the value of "everyone" though. Spidering (for legit purposes and otherwise) is mostly just background noise at present; the real bad actors - cyber criminals - already ignore robots.txt, and not every good actor would significantly benefit from ignoring robots.txt. The only real reasons a good actor might have for ignoring it are for better archiving (as with IA's proposals) or more complete search engine indicies, but if the reason for the content being excluded via robots.txt is that it is highly dynamic, transient, or just fodder for bad robots, then it's of minimal value to search engines anyway. Even if some (or all) of the search engines were to follow IA's lead on this, I think they'd still be looking at balancing that with more intelligence in their spidering just to avoid the risk of cluttering up their databases with broken links and expired data, and that's likely to limit the bandwidth requirements considerably.