Slashdot Mirror
Microsoft Watching What You Watch
Arkham writes "According to this Wired article, Microsoft
has contracted with a company called Predictive Networks to track the viewing habits of Microsoft TV devices. The Predictive software creates a "Digital Silhouette" that is described as being able "to tell them that Joe watches a lot of baseball, likes Situation Comedies, and
responds favorably to commercials that use humor."." I've always said
that I'm cool with my Tivo tracking what I watch, provided it never tells
anyone my name and address to anyone.
If it meant I watched more
targetted advertisements, I'd fast forward less.
I think it's a given that any set-top box that you buy today (TiVo, ReplayTV, anything by Microsoft) is likely going to track viewing habits and use that data for market research. I'm not 100% against the concept--if it means that there will be more shows that I actually like, I think it's a good thing. What we have to worry about is when the media line is crossed, where the data is used to target you for direct mailings, telemarketers, spam even.
Imagine you are watching TV, and you watch a lot of National Geographic. Suddenly, you find yourself getting magazine subscription requests in the mail, telephone calls from NG about becoming a member, and e-mail in your Inbox about the Web site, all just from watching TV. This is something we need to remain vigilant about, that the companies don't use the data they collect in an all-out attempt to sell us their wares (no pun intended).
Karma: Excellent Birds (mostly as a result of listening to Laurie Anderson)
On Sky Digital TV in the UK, there are advertisments that carry an on screen reminder to 'click now' for a free sample. Clicking takes you to an advertiser's page that can then connect through your phone line to send your address details.
To step up to 'buy now' functions isn't that much of a stretch of the imagination.
The Predictive software creates a "Digital Silhouette" that is described as being able "to tell them that Joe watches a lot of baseball, likes Situation Comedies, and responds favorably to commercials that use humor/
How in the hell does it know how he responds to the humorous commercials? Hopefully they just know that he doesn't change the channel away from them, and that they don't actually know his response somehow....
~ now you know
Perhaps it's able to tell whether you FUHFUHWD through the ads or not. Or if you go back and rewatch an ad you've already seen.
I wouldn't like my viewing habits being documented even if it was Tux the Almighty doing it...
PageTurner Reader: open-source e-reader for Android with cloudsync. http://pageturner-reader.org
> So how does it intend to find out that Joe responds to commercials?
By the surprisingly easy method of channel monitoring. It's been shown time and again that if a commercial is uninteresting, people will surf around to other channels, then come back when they think the show is back on. This behavior is also what drives the short-long method, where the first commercial break in a show is short (two or three ads) then the next is longer, and the network can charge more for the ones in the middle of the long set, since they have a higher "hit position" than the ones in the beginning. So, the device simply monitors which commercials keep Joe from changing the channel, and then looks for trends in those commercials to see what themes keep his hands off the remote.
And you thought this was simple? Networks spend millions learning stuff like this.
Virg
The more I watch the news, especially since September 11, the more I realize there is very little new information dispersed to the masses. Instead I see people waving signs, but no mention is given to their history or how those signs were written in English. News, more and more I think, just tells us mostly what we want to hear or think we would hear.
It seems to me that one of the primary purposes of advertising is to sell you things you didn't know you needed. So if advertising is so targeted that the commercials and products only reflect ones already forged tastes, then how does that help to sell more widgets? Like the news, this sounds like a way to sell us what we already know about or what we already want, and doesn't seem to lend itself to increasing sales or opening new markets.
I can't get too worked up about this... as long as the consumer knows when he buys a Microsoft TV product that it 'comes with' this kind of monitoring. That, to me, is the key -- full and open disclosure, and a consumer educated enough to know what that means.
Really, if you have a Yahoo 'home page' configured, you're already providing information about your preferences -- voluntarily -- albeit on a lesser scale then what MS TV will do.
If you use one of those 'shopper discount' grocery store cards, you're also providing this kind of information, in even greater detail. If you purchased a pregnancy test or jock-itch ointment last week, it's in a database somewhere if you use one of those cards, and the fact that they don't individually target you NOW for marketing based on this information doesn't mean they won't in the future.
From the above article: "...61 percent of retailers surveyed either have or plan to have frequent-shopper programs. Already, more than a quarter of all supermarket sales are tracked with the cards."
That shopper discount card sounds much like what MS TV plans:
"Scott Oddo, director of research at Predictive Networks, said the collected information does not connect viewers' interests to their names or other personally identifiable information."
Anyways, I'm sure all this info that they supposedly need is certainly helpful to them as a company, but I think those surveys they send to your house are more effective than this. Besides, you're opening up a whole can of crackers who will want to break in and steal all the data on their neighbors that they can from MS's database, just to say they can. And if they're really malicious, they'll know exactly when you're watching TV, and when you're not, and can then rob you.
This paranoid delusional tirade has been brought to you by the number '5', and the letter 'F'. :-P
First of all, it is NOT OK with me, TiVo doing it, or anyone else.
Second of all, MS stuff is NOT 50X better, by any standard. It is not better at all.
Third, there is nothing irrational about an extreme dislike for Microsoft. You can remain blind to the subject as much as you want; but this is about far more than just an inferior product. And most frightening of all is that they quite sincerely don't see anything wrong with their behavior. So it's your responsibility
and mine to object in any way you can. Hey, I have to use MS stuff too from time to time; but I try really hard to keep it to a minimum. It's the least any of us can do.
Oops
And this is why this technology is BAD!!! Personalization is not about pigeonholing a person. This is why personalization has not worked for anyone. Companies think that just because a person likes such and such then they must also like that. Well beep, wrong answer.
The problem with personalization technology is that it works on past behavior. And the thing is that we humans tend to be quirky and fad based. As a result it is basically impossible to tell what people want. Hence why did things like Rubik cubes, Beenie dolls, Cabbage Patch kids cannot be predicted.
What is harder and companies are not solving is contextual personalization. Basically what happens is that instead of attempting to figure out what the system thinks you want, the system helps you figure out what you want.
For example lets say you want to buy a stero. Well you goto store 1 and get quotes for pieces A and B. Then Store 2 get quotes for piece A so long as you buy C. Get an agent to optimize the situation.
How could I see this in TV? Simple, lets say that I am cleaning the house and feel blue. At that point the "agent" should set the channel to something like "The Country Music Channel". This is a system that I would like!!! But it is very different...
This is basic agent technology and very difficult to do.
"You can't make a race horse of a pig"
"No," said Samuel, "but you can make very fast pig"
Next thing you know, my 79 cent loaf costs $1.39 and I'm supposed to feel lucky when they sometimes offer a special membership price of $1.10. Uh huh.
If we assume that:
A) every company is out there to maximize their profits, and that the best way to maximize those profits is to balance margin (markup) with # of units sold,
and
B) in a proper implementation of such "tracking", every company in a given industry (in this example, a grocery store / supermarket would track ALL products, not just those of a certain brand) would take advantage of such technology,
then one would assume that the technology would be used to find the "best" pricepoint for every product. Best being defined as the point at which raising the price hurts unit sales to the point that net profits go down, and lowering the price does not increase the nunmber of units sold, but detracts from profits. A BBA-type can probably tell us the proper terminology for this, but I think its a simple concept.
Let us assume that, on our way to buy groceries tomorrow morning, instead of walking into our normal grocery store, we walked into a store with a solid tracking system. Every item in the store would be effectively priced and positioned - nothing would be "overpriced" for the effects of overpricing would be noted and corrected for. True, nothing would be underpriced either, but (and here's the contentious part of my post)... would we really be paying more on our grocery bill? How many times do you look at a product and think to yourself "if this was x dollars cheaper, they would sell TONS of these!"
Of course, if the grocery store sells only one brand of stuff, and/or there is only one grocery store in town, (ie monopoly) then the whole theory gets ugly, quickly. But assuming that grocery stores remain grocery stores, and they compete, and carry many different brands, then tracking would produce a net benefit for the consumer.
Of course, all of this tracking (grocery store concept) doesn't really need to be tied to an individial identity, but there are other benefits to that. All I am suggesting is that shopping at a place where our purchases are tracked has the potential, though healthy competition, to serve us in the long run, by killing off the inferior products and effectively pricing the good ones.
I just wish some of the restaurants around here (I live in Antigua, in the Caribbean) would grasp the concept that, in times of tourism downturn, restaurants would grasp that if they didnt charge $30USD for their entrees, they wouldn't sit empty, and that lowering their prices would result in an INCREASE in profits... but I digress)
-- "Ignorance more frequently begets confidence than does knowledge." (Charles Darwin)
Hmm, if that's true, explain to me how the price of a loaf of bread at this chain has a different price at each outlet in each city. Hmmm, the same chain stores 5 miles apart charge different prices. Hmmmmm. They know the demographics of the people coming into each individual store and they know their specific spending habits at the stores. Hmmmmmm.
No duh. I don't think they care about me specifically. They track us as a group and determine how to manipulate prices that way. I just refuse to be part of the group being tracked.
It's not so much that it's been done before, it's being done by one that is troubling. Then you get data mining where all those innocent little bits of information about you are collected, analyzed, and determinations made about you.
My guess is, the only ones left defending Microsoft at that point will be the Microsoft plants. I wonder how close we are to that being the case now...
Hi!
You may well have seen a grocery store chain experiment with prices--that's a practice that's been going on for decades. (Grocery stores routinely change different prices in different neighborhoods--a few cents more in rich neighborhoods (where consumers will pay more for convenience) and a few cents less in poor neighborhoods (where consumers may be more price-sensitive). The "frequent shopper" card systems won't really impact that--grocery stores can map price/demand (elasticity) curves already.
What the frequent shopper programs do is help the stores manage what products get displayed, how they get displayed, and how they are priced. They do this by identifying the "core customers"--both of the chain, and of that particular store.
Most consumers purchase most of their groceries at one particular store. They might shop at a different store (it's adjacent to your child's school, so you can stop in on your way to work) occasionally, but that big hundred-bucks-a-week trip usually happens the same place. By offering lower prices on selected items, the stores entice the frequent shopper to sign up for the card, and permit his shopping habits to be tracked. While this offers the theoretical possibility of monitoring the customer (we get our prescription drugs at the grocery store--if I start getting scrips filled for AZT, does that mean I have AIDS?), it offers the immediate opportunity of selling coupons to advertisers. (To wit: I buy dog food--even when I am not buying dog food in that particular trip, I almost always get a dog food coupon at the register.)
The real advantage of a frequent shopping card, though, is identifying the buying habits--in the aggregate--of the store's core buyers. It helps enormously in making "plan-o-gram" decisions: how much of what to stock where. Example: last week the deli ran out of salmon four days running. Should we increase our daily order of salmon? Well--if our data shows that most of that salmon was bought by frequent shoppers, the answer is obviously yes--these are customers who will likely be back for more seafood. On the other hand, if very few of our frequent shoppers bought that salmon, it might be wise to wait--we may have had a statistical cluster of salmon-swallowing tourists in the neighborhood. In a similar way, we can identify whether our core customers buy more of our store brands or the name brands for particular products. We may find differences between this behavior in different stores: in stores where our brands do better, we give those brands more space; where our core customers prefer the name brands, we give the name brands more shelf space. In any case, we tailor the shelf space in each store to focus on the product mix favored by the frequent shoppers in that store--that may mean more salmon in some stores, and more produce in others. (Real live example: there is a chain grocery store in Morrisville, Vermont--a tiny town thirty miles from the Canadian border--that has five different varieties of fresh mushrooms in the produce section on any given day. Why? Because their core customers like mushrooms. [Real Vermonters might suggest that this store caters to quiche-eating flatlanders, and offer this as proof, but I digress....)
In the example that you cite:
The store might test different prices to determine your resistance to a price increase (this is called "elasticity" by economists--elasticity is to Econ majors as pointers are to CS majors: if you don't get the concept, you tend to go find another major). If you're going to buy French bread, and you're willing to pay $1.39, that's the price. The frequent shopper cards may help in letting the store measure price resistance among the core shoppers (that is, if 80% of the store's french bread is sold to frequent shopper cardholders, and they demonstrate a near-horizontal elasticity curve [change the price, they don't care] then the store can safely hike the price of french bread). But stores have measured elasticity like this, as I wrote above, for decades--all the frequent shopper cards do is let them measure price resistance more accurately.
John Murdoch
In related news, Starbucks's coffee card, one likely gift to be in your stocking, comes with an interesting caveat... if you want any value-add to your card, you have to tie it to a Microsoft .NET Passport account.
All your coffee are belong to us.
Slashdot: Everything in Moderation, including Moderation itself.
Suppose I try to watch Futurama but a ballgame is on, or some post game show, neither of which interest me. What do I do? Generally I turn the sound low (but not muted) and ignore the TV until the stuff I don't care about is over, and then watch what I tuned in for. Or I end up turning the set off in disgust if Futurama isn't aired at all.
So, what does the profiler say? That I like post game shows? No, I most certainly do not like them.
I'm a little concerned about data collection and targetted advertising, but I'm more concerned about the data being wrong or misinterpreted and ending up with more of the stuff I didn't want in the first place.
Marketing would dearly love to know what you watch and what buttons you click to get there and when you change channels and all of that. For at least the large majority of subscribers, they certainly don't know it now. Cable network ratings come from the same Nielson household monitoring service that over-the-air ratings do.
As noted, if you hook the coax straight into the TV, or if you have a typical analog cable box, there is no return path for usage data, even if the devices were collecting it, which they are not.
If you have a current generation digital box there is a return path, but none of (a) the software in the box, (b) the bandwidth available in the return path or (c) the systems receiving data from the STB is geared to handle the volume of data that channel changes and viewing patterns would require be transmitted.
Advanced STBs with built-in cable modems may be capable of recording and transmitting such data in the future. At least some of us technology folks at the cable company are worried about the privacy concerns of doing so.