With Metcalfe's law, bitcoin is indeed quite promising, as the users have been increasing and are expected to increase further.
This means the value of bitcoin is increasing. Could this argument be valid?
We do not know whether the number of "bitcoin users" is increasing.The sources that I know (such as blockchan.info) do not give that information. They give some quantities (such as wallet software downloads, transactions per day, total BTC volume per day) from which some people claim to be able to derive the number of users. However, those quantities include an unknown amount of operations that do not imply real additional use. Some of them (like total BTC volume) have been relatively constant for the last 6 months. Moreover, there is no information at all about people who stopped "using" bitcoin, e.g. after buying a bit just for curiosity.
Yes, we actually do.
blockchain.info stats aren't the only source saying so. Online wallet stats, number of vendors accepting Bitcoin... there's plenty of evidence of growth well above linear increase (cue: "doesn't mean it's long-term exponential").
Your argument runs down to: but all those stats could be faked/manipulated by entities high enough in the decision chain of exchanges or websites. While theoretically
possible, you have to ask yourself it is the most
likely explanation of the data.
There's plenty of fraud and deception in the Bitcoin ecosystem, but what you seem to have in mind is a level of organized deception that you cannot simply "claim" to be the reason for the data. If you have solid proof for it, let us know. Otherwise, I could dismiss anthropogenic global warming by claiming that measuring stations world wide were manipulated by Jewish space lizards -- it's a possibility, y'know.
tl;dr The most likely reason to generate data indicating exponential growth in usage is actual usage. Claiming otherwise requires evidence, which I have yet to see.
It doesn't have to be manipulation. Say you're laundering coins on a regular basis. Say you're using Blockchain.info to do this. How many throwaway wallets a month would an automated system built for this purpose generate? What about several of them?