[Aramgutang]

While I agree with your point, a part of me can't help but wonder about the possibilities that would open up if we had the capacity to extract enough semantic information from a photo:

If I'm looking at my old holiday snaps and reliving the good times I had, I might be tempted by an ad for cheap flights to a similar destination.

If I'm looking at recent photos of my relatives' adorable children, I might be tempted by an ad for children's toys to send them.

If I'm looking at a friend's photos from an event they attended and enjoyed, I might be tempted by an ad selling tickets to that or similar event.

If I'm looking at a photo of myself from a time when I was in better physical shape, I might be tempted by an ad for a gym membership.

etc...

I doubt Facebook will ever be able to get to a point where they can make predictions as accurate at that, but given the trove of personal data and image analysis algorithms they have, if anyone could do it, it'd be them.

[JonnieCache]

The key word there is "might."

FB and its believers think that the chances are high enough to make big money, or at least that they can make them high enough.

In actuality they aren't. You might think about booking the flight, but the chances of you actually reaching for the credit card based on idly looking at photos are too small to do worthwhile targeted ads.

The idea that FB can just tune their algorithms until they know exactly what you're thinking is a fantasy based on...

1. an overly optimistic estimation of the SNR ratio of the info people put into FB, and

2. on the hilarious fallacies of the aforementioned Market Economics Fairy which assumes that everyone is always a rational actor, and that everything people do has definite reasons behind it which can be traced and understood, and their future behaviour predicted from, and crucially that these reasons can be modelled with traditional reductionism. This is plainly not true.

Admittedly my opinions here are based on guesswork, just like FBs are, but I contend that I have the benefit of drawing my conclusions without a hundred billion dollars worth of wishful thinking clouding my vision.

[a3d6g2f7]

"everyone is a rational actor"

And that's why the message to Harvard graduates[1] (purportedly America's brightest rational minds) makes deliberate reference to a man[2] who has consistently proven this to be false. The message also is centered around another pillar of rationality: the idea of "luck".

1. A link to this was recently posted here on HN.

2. One who was "lucky" enough to be awarded a Nobel Prize.

[mrvir]

Even if they could do all that it is still wild guesswork with too many possibilies. And the effectity of that fantasy also assumes that there is endless supply of advertisers from all different industries. The reality is that if, for example, only some dating sites and hosting companies are willing to pay for ads that match your profile then it is only dating and server hosting ads that you will see.

[a3d6g2f7]

This is the trap. This is what FB employees keep telling themselves everyday. This has to work. It just makes sense.

You can keep saying the data is valuable. No one is going to question that it is. We all have imaginations. And advertisers will be curious. They will try advertising on FB.

But there's just a big difference in human behavior in the two examples I gave. They are different. Not because I can make argument based on reason for the difference. But because it's what the facts show. There is a long history to this that predates the web. You might say it's been thoroughly tested.

So, you can make all the arguments you want that people should behave a certain way, and maybe you draw on examples of your own behavior or that of others you know, but when it comes down to making money and growing like Google does, it matters more what people actually do in real life. As opposed to how we think they should behave.

Yellow pages and photo-albums are things that have been around for a very long time, long before the web. If people looking at photos (FB) are at the same time looking for products and services, then why didn't photo-albums before the web have heaps of advertising in them? The answer is because there was a different book where it was more appropriate to advertise.

Read the article on the MIT or Harvard blog again. What you are describing, especially a tactic like image analysis, is exactly the type of approach he is saying is just too invasive. This kind of approach to advertising removes anonymity and violates people's privacy when it's counterintuitive to do so, according to the article.

[alttab]

That last bit Proves that Facebook truly is the work of one man.