I don't agree with you that people should avoid making any type of prediction. I think a reasonable prediction is that investing 2 TH when the network is currently 10 TH would yield somewhere between 16.7% and 20% of hashing power.
Anyways, it isn't worthwhile to discuss future events with someone whose only position is that no one knows what the future holds.
i disagree with you that its a reasonable prediction that we will see ANY drop off in hashing power. as i said, there are just too many variables. how does your prediction take into account potentials like a nice increase in bitcoin value (which historically shows a correlating increase in network hashrate)? in say butterfly labs shipping singles and mini-rigs in quantity? in potential new companies sprouting up delivering economical mining machines? or the numerous other potential events that are realistic?
your forecast is about as valid as a professional weatherman trying to predict the weather on a single day 6 months out. just too many variables, and the further out you try to forecast, the less valid your predictions.
maybe we see a drop off, maybe we see relative stability, maybe we see a network power increase. why you think you have special insight and skills of prognosis professional forecasters repeatedly show not to be in possession of, is either overconfident or delusional.