On the question of Online subscription pricing. There are additional points to consider.

Price elasticity is based on a market of competitors… which has to be defined. Since there are no geographic boundaries online- substitutes and complements are a little fuzzy.
The thought about segmented pricing, is adopted by several information services already – general content is free, it is the premium articles and videos that require subscription.

Question to be considered: can the information be obtained in other news room/ blogs/ forums.
Since one user can copy the article and forward it to their network, the “Value” is subject to free-rider complexities.

So, the web analytics needs to map the click through process to identify, this “market”. That analysis will allow us to segment customers that go for the news only category, news + blogs+ discussion groups, news and competing news agencies. Topics of interest would also be important – since that adds to the layer of segmentation.

Finally, I like the approach you took on the demand curves… I have used similar processes to estimate demand and value components – it is not just data mining but a significant effort to translate ‘expert testimonials’ to get to the ‘ right average’.

One of the issues we should keep in mind while talking about ‘Demand curves’ is that not all products and services display a typical demand curve – one that is displayed as an ‘inverse demand function’. When I was doing pricing analytics for a somewhat unusual and relatively unknown industry (Highway tolls), we discovered how ‘utility’ can be such a powerful factor in determining the demand curve. The utility function obviously depended on a host of factors like substitute routes on a transportation network, time of day, demographic characteristics etc. As a result of these factors, the resulting demand curve turned out to be fairly inelastic with respect to price – until a point. And after this threshold price-point, the demand fell precipitously.

To exemplify, the network demand for a certain tolled road may vary within 2000-2200 cars per hour for a toll ranging from $0.10 to $0.50. However, as we raised the toll to $0.55, the demand would fall to say, 1500 cars per hour.

A scenario like this may prove true to the specific example of NY Times. Depending on their pricing structure (for example, NYT may determine that price to access video content may be different from accessing editorials and blogs or the price may even vary with geography or socio-economic factors thus pandering to varying utilities), they may find that the demand function may not be as predictable. With a little market research and limited release environments, they may get some idea of the utility function and zero in on a starting price point(s). But such a price point may have to be played around with a few times before figuring out a longer term pricing strategy. As happens with dealing with most inelastic and ‘abnormal’ demand curves, NYT has to be ready to capture and manage data very well which, with the right processing, can give them tremendous insights into readership which may even lay the foundation for a non-free news world.