Not to mention service oriented designs that allow you to charge for the service, not the UI front-end. But I will leave that out of this discussion.
Common to all models is the process of writing the app, publishing it on the AppStore, and then advertising it so that people know it is out there. There is little to no problem if the app is unique (or offers unique services) and if the advertising has been excellent. Positive reviews from trusted people (friends, prestigious bloggers, magazines) boost your sales. But what if your app is not unique or if the advertising has not been to the sky or if reviewers just didn't have the time to try your app out? What if you app is not unique but just better (or so you hope) than the 15 others out there? How do you make that first sale? How do you get that first positive review? How do you get that tweet or Facebook like?
Pause. Haven't we solved this problem already? Fast rewind 10 years and we are in the era where the PC is gaining ground. Thousands of applications are being written and published on the web or discs. They are being advertised on the nascent web or popular PC magazines. There are many similar apps that offer more or less the same functionality but each has a specific way of doing it. How did we then get the user to buy our application in those days? And as most people would remember, back in those days (as is still the case today), quite a lot of paid software also came in a trial edition. There were (and are) two distinct kinds of trial software:
Time limited editions: Expires in 29/30 days. To continue using please buy a serial key or download the full version by clicking here.
Limited functionality editions: Free but comes with teasers --disabled menu items or blocking pop-ups that "encourage" the user to buy the "premium" version.
And I would say that worked out pretty well. So well that when web-services came along they adopted this model and gave away free access (limited by time or functionality). For example, many cloud "drives" give you X GB free and an option to get more by paying [more]. Some photo sharing/editing services give you a basic account and a "pro" account. If you like it, buy the premium content. The only downside to this, in the PC market at least, is that if people don't like the trial version you do not get the sales. That can mean taking a small loss because your investment in trial version would go waste. Users may not be so willing to try your later inventions too --once bitten.
Rob mentions that the 1st model has been by far the most successful. But it does not mean that the trial edition model won't get you those dollars. Quite to the contrary, given a choice to try it out, most people will try it out. Given an option to not try it out, they may not risk buying the wrong app --assuming there is a competition. Nonetheless, there are some differences between the PC/services market and the AppStore that support Rob's observation. And here is where I enter the realm of theory and speculation.
The Promise: In the PC market, you do not have an AppStore (well, leaving Mac AppStore out, it is not yet clear how it would affect consumer behaviour; am also leaving the repository market that most GNU/Linux systems come with for obvious reasons) --you have a free and open market where anyone can sell anything they want to. The biggest downside to this free and open market is the lack of trust. Unless you are a big professional/prestigious company, few people would trust that your software would be of high quality, will do what you told them it would do, and won't steal their data or do weird things to their computers. Priced software usually comes with a good reputation --if it is expensive, it must be good --but note the word "usually". In the AppStore world, Apple's policy on design, app advertisement (not iAd but your app's home page), developer identity verifications, membership fees, and programming has ensured that most users (even those who are not aware of such policies) have come to expect that apps will do what the developers say they will and that they will be of good (if not high) quality.
The cost: The biggest inhibition to buying something is the risk that you will make the wrong decision. This risk is measured by the cost of the product and the cost of using the wrong product. Buying a free virus (labeled as time management software) has a higher cost than buying a $4 coffee that didn't taste well. I have covered the first type of cost in #1. So, let's focus on #2 now. It is conventional wisdom in the FMCG retail world that brand shifts happen most at the lowest priced products. Most companies would enter the market with low priced or miniature versions of their products. Consumers aren't afraid of trying out something new if it does not cost much. Trying out a $15 tube of lotion is more risky than trying out a $2 soap. Many companies enter this market with a $2 soap and lure the customer into the promise (#1) --then the customer is willing to try that $15 tube of lotion. The point is not brand loyalty but that customers are more willing to experiment (and lose the occasional buck) for a $2 product than they are for a $15 product. Barring a handful of the apps that are priced in double/triple digits, most apps are $0.99 - $2.99 on the AppStore. For someone who buys $400 phones, that may not be too much. Sure, the customer will regret losing that $0.99 (and probably won't trust your company again) but not that much.
Now that the promise has been made and the cost of getting it wrong is not so high you will see that the 1st model can get quite a lot of buyers. An impartial review, right where you make the purchase, adds to credibility and reinforces the purchase/reject decision. Of course, it is common sense to think that the 2nd model (and I do believe) will get more buyers but it also costs more to produce and maintain two versions. Plus, you risk negative reputation with a poor trial version feedback --which is nominal because it applies to the paid version too. But at least with the paid version you recover some of your cost.
And hence, in the AppStore world, most developers initially went for the 1st model --it allows them to recover more cost (including that $99/year) than the 2nd model allows. We can see a growing trend in the ad supported trial version model --a combination of 2nd and 3rd models. Developers/companies who can get the two version cost tamed (e.g., by better software design and support model) can make more money with the 2nd model.