The curious semantics of "up to"

There's been a lot of discussion in the UK about the appropriateness of using the "up to" tag in broadband marketing campaign connection speed claims. Given that average speeds tend to be significantly lower, maybe these would be a better basis, but this would likely differ from exchange to exchange, which doesn't make for for a tidy national advertising campaign - and obviously, average speeds are less flattering to some than others. It's a genuine quandary, and doesn't bolster customer faith in the broadband value proposition. Still, in the UK, we're talking about missing expectations by a few megabits, though these can be significant - see this study previously pointed out by Benoit, and locate the UK in the rankings. In the relatively fiber-rich French market, the stakes in the "jusqu'à" marketing war may arguably be higher, partially because UK consumers are used to disappointment [:-)], and partially because the gap between claim and reality will yawn wider in future. If I'm reading this article (link removed - see UPDATE below) correctly, a test of Numericable's "Up to 100Mbps" FTTx product by an author previously on the 30Mbps shows effectively no change in throughput - a gap of over 7oMbps. Now that's marketing!

UPDATE on 7th October: It turns out that the PC INpact guys were running the test using a plain vanilla DOCSIS 2.0 modem, which accounts for the 30Mbps peak throughput. They have issued a mea culpa, and are apparently awaiting a "pre-3.0" NetGear CBVG834G to replace it.