You raise a good point: the post is just one anecdote. Yet, if you read the post, you have to admit it was thought-provoking. How to explain this contradiction?

The reason, I suspect, is that a survey may average out all the trends and observations about specific use cases for services. You start with colors, you end up with gray-brown.

Of course, billions of dollars in valuation are at stake. Facebook and Twitter among others were built very fast and are probably still open to disruption if they are not watching these trends. People are jumpy.

> You raise a good point: the post is just one anecdote. Yet, if you read the post, you have to admit it was thought-provoking. How to explain this contradiction?

No, it wasn't. I personally don't like sandals. I'd happily give you lots of reasonable-sounding explanations for why I don't, but it's just the preferences of a single guy in the end. Nobody would spend more than a half-second considering whether my personal choice is relevant to the futures of the footware industry.

Yet suddenly because it's the internet and teens, we think the subjective personal experiences of a couple teens is more important than actual data. It's ludicrous, and not at all thought-provoking (except in the sense that humanity's continued inability to think statistically is provoking).

> I personally don't like sandals. I'd happily give you lots of reasonable-sounding explanations for why I don't [...] Nobody [considers] my personal choice is relevant to the futures of the footware (sic) industry.

Of course it is relevant. In your case, the footwear industry would like to know whether the general public likes sandals or not, in order to decide whether to make more or less of them, and what types of advertising should be purchased. We can go with the initial assumption that there is nothing special about you, as a sample, and work on the principle that you are representative of the footwear buying public as a whole. So, we now know that not everyone likes sandals, and since we know other people that also hold the same view, we can be confident that you are not absolutely unique. Of course we want to seek out more opinions, and determine more accurately how widespread this dislike of sandals is, but I would be foolish to completely ignore this data point.

Similarly for teens and Internet services, the single data point here is interesting and thought-provoking because we know it isn't a freak outlier, and is instead somewhat representative of teens as a group. Since the actual quantitive information (that is, user retention and engagement at Facebook, Snapchat et al) tells us only what is happening, we must seek out qualitative opinion pieces like this to determine why. Everyone here is capable of understanding the limitations of small sample sizes and self-selection effects, and should be able to control for that. And of course, it's also just interesting to know what ones fellow human beings feel about a subject, and even important when there are business models predicated on these same humans doing things because their friends are. Network effects like that are obviously one of the main drivers for social network adoption, and thus increasing the value of a product.

> We can go with the initial assumption that there is nothing special about you, as a sample, and work on the principle that you are representative of the footwear buying public as a whole.

That's an absolutely terrible assumption. A sample of one is never ever representative.

A sample of one is more likely to be representative than an outlier, although I admit I'm making assumptions here about the underlying distribution.

This is a blog post equivalent of qualitative user research, which, in addition to quantitative methods, is really useful to understand why people are using technologies the way they do. Research and understanding usually starts from small samples and anecdotes and expands from there via more rigorous methods towards the truth.

Also, the teenager polled was asked about apps and services they use socially which actually expands the pool of data considerably.

This isn't "what is your favorite ice cream" which is a purely personal-singular choice. This was "what do you and your friends do" which expands greatly the number of responses.