Forecasting phone sales is relatively easy. We know roughly how many people with mobile phones there are on earth (3.2bn), how fast that is growing, how often they replace their phones (a little under every two years) and how that is changing. So, you can do a five year forecast with a pretty high degree of confidence that you won't be too far wrong.
Forecasting smartphone sales is then a subset of that. You know how many phones sold now are smart (just under 50% in Q4 2012), you know how fast that is changing and roughly how many of those 3.2bn people cannot afford to pay more than, say, $50 for a phone, and assume that in 3-5 years all of the rest will be buying smart phones. There's a little more variability in those numbers, but again, you're not going to be very far away from the right answer.
Tablets are a different matter. Three years ago people though it was a 5-10m-unit a year market - Apple sold 65.7m in 2012, and Chinese OEM/ODMs sold tens of millions more. Well over 100m will be sold in 2013. But where is the ceiling? Three years in, we're a little adrift. The line points up to the right, as it does for smartphones, but for smartphones we have a good idea where it will stop: with tablets we don't.
Which of these proxies is most relevant? We can only speculate. PCs are an obvious match, but are bought only every 5 years and are a 'one per household' product, neither of which is true of tablets. Tablets themselves, as the definitions blur, start to cannibalise, not just PCs, but also smartphones, especially high end ones. An iPad Mini plus (to give an extreme example) Nokia's 105, which costs €15 and has a 35 day standby time, might be a good combination for some, while people in Asia do make phone calls from 7" tablets. And then 'tablet' itself encompasses devices priced from $75 to $600, which don't really compete with each other.
To me, though, the most interesting benchmark is not PC sales, which is where people tend to look to see what tablet sales might be, but more towards the right of the chart. After all, that's the number that tells you how many people have $100-$300 to spend on a magic box connected to the internet. Maybe they'll all buy a big one and a small one. In that case, tablet unit sales might be 10x what they are now.