Advice is the world's cheapest commodity. To avoid comparison with advice, the following are offered as gifts.
There are many ways to achieve "stickiness", a property that promotes regular returns on the part of visitors, in a web site. One of them is to allow people to discover what they like (they generally don't), and to watch their own characterizations of what they like change. More to come.
In a hierarchical organization, the probability of an analysis yielding a "deep" (important and nonintuitive) result that is implemented in the organization is inversely proportional to both the depth of the result and its importance to the organization. More to come.
In the late 1960s, during the inglorious reign of the kind of large-scale modeling of physical/economic systems that gave a bad name to operations analysis, I was often charged with projecting the value of a certain probability, regarded as a figure of merit, for competing system designs. I became rather good at designing and building computer simulations for estimating this probability in a simulated environment of interest. Management's appetite (and many times, my own) for greater and greater levels of detail in the modeling was very high; the added detail lent an air of depth and "confidence" to the enterprise. The added detail, though, meant that more and more random variables, many with large or totally unknown variance, had to be introduced. The result was not comforting to those looking for better answers. Implications to the modeling of consumer behavior are equally striking. More to come.