Strange Loops - Blog Archive: November 2006
Strange Loops Journal Archive: November 2006

Blog || Politics || Philosophy || Science || Fiction || Quotes

DNA Databases, Privacy, Targeted Advertising
November 16, 2006
Permalink ||

Currently at least six states plus the federal government require DNA sampling before criminal conviction. One of those states just overturned the law. The Minnesota Court of Appeals ruled that the state law establishing a DNA database and requiring a biological specimen for DNA testing was in fact unconstitutional. It violates the fourth amendment, the appeals court declared.

Which likely means that the Supreme Court will have to step in and look at this issue before too long, especially if the law is challenged in other states (which is likely now that precedent is set).

Why should anyone care? Well, aside from the plain old creepiness and annoyance factor of having your DNA analyzed and put in a big government database when you are accused of a crime -- even one you did not commit and are not convicted for -- there's the question of how that information will be used.

For one thing, can the database's information integrity be assured? If mistakes are made, will people be able to find out and correct it? Other government databases suffer from this problem: for example, no-fly lists have mistakenly grounded completely normal, innocent people with names which might coincidentally be roughly similar to those entered into the database as possible potential suspected terrorists (based on no particular standards of evidence). Since there is no way to check your own status in such a database, nor any process to correct such errors, the database's information is next to useless. Yet it still gets used, and still ruins the plans of many a traveler trying to go on vacation, get home to family or go to a business conference.

The same deal is likely to happen with any database where proper care isn't taken to ensure the integrity and provide for catching and correcting errors. The problem is, such databases are historically used heavily as fact even after holes are demonstrated in their integrity. So having these huge DNA databases can lead to some major problems.

What happens when the database gets used for things other than its original intention (which naturally happens with such an apparently valuable resource)? What happens if, later on, people propose laws restricting the marriage of two people whose genetic code would lead to critically disabled children -- you have to consult the government database before you can get married. What happens if, at some point, private companies contract out access to the database for other purposes, at first maybe to help verify identity for banking transactions, or to track down biological parents for adoptees. Later on, insurance companies will use this information to set your rates -- sorry, you have to pay a much higher premium if you carry a particular allele at gene locus M314LQ6 on chromosome 13 because it has been correlated with a slightly higher risk of dying of cardiovascular complications. The database says your DNA has this allele, so you pay more. Can't argue with the database.

Honestly, I think this sort of privacy invasion seems inevitable. And it comes from much more than information databases, or using our DNA Gattaca-style (this will, I guarantee, happen within our lifetimes). It will come from RFID chips (radio-frequency identification), the tiny ultra-cheap little chips that broadcast out a signal when pinged by a reader. They're used right now to help track inventory in warehouses, to track high-value, easy-to-steal items in stores (Walmart put them on razors), in ID cards granting access to restricted buildings, etc. They will soon be mandatory on state-issued ID in Britain, and it won't be long before we follow here in the states.

All this despite evidence that they are horribly insecure, that someone can steal the data on such a chip from a distance and have access to all the data, or manipulate it to their advantage. Yes, it's handy to be able to just swipe your credit card near something (or walk through a scanner) and have your groceries deducted without having to push any buttons. Yes, it's handy for companies to be able to track items in a store -- or potentially after they have left the store (the originating company aims to create a global satellite tracked database of every item containing such a chip). But it also means other people can just swipe your credit card information by passing within a few feet of you (or from across a room or street, given a decent reader). Identity theft becomes easier, despite this technology being proposed as a security measure. Are we safer from terrorism with these sort of IDs issued by the state, when anyone can hack and alter them with unbelievable ease?

And getting back to the idea of tracking every item purchased at a store, even after it has left the store. Think of all the data that companies -- clothing stores, grocery stores, whoever -- will be able to accumulate on their customers' habits and item use given this technology. They will be able to target advertising much more directly -- personalized advertising, assaulting you with ads for new jeans the day your old jeans wear out, harassing you with ads for mustard because the one in your fridge is getting a little low. If you don't think this is possible, look at what grocery and other stores do now to track your purchasing habits -- membership cards with sales prices only available to those with a card from the store, so that they can tell exactly what you buy (note banks keep track of all non-cash purchases, and some have argued cash money will go away before long).

Targeted advertising has proven itself effective over the internet market: cookies track what city you live in, what websites you visit, what you buy online, what Amazon items you've browsed recently, the words scanned from your Gmail messages, and allow targeted banner and text ads based on your history. It's huge right now in cyberspace marketing, and it's going to grow here in meatspace very quickly.

Frankly, the privacy we are used to is going to disappear. Quickly, in a historical sense, but slow enough in a modern everything-changes-fast sense that we will get used to it in new fads and new techno-gadgets before we realize the long-term effects we've started creating. The world of the future -- and I'm talking still during our own lifetime -- will be drastically different, I think, insofar as privacy goes. And yet most people won't notice, because they'll be getting used to these things in little increments. Soon people will think it downright odd why anyone would want or expect to keep private the items they've purchased (and from where and for how much), or their medical history, or where their car travels (GPS tracking combined with a database -- you know, so insurance companies and cops can verify whether or not your vehicle was traveling the speed limit when it got in that traffic accident).

Can't stop progress, whatever that is. Just interesting to observe it as it approaches, and speculate on how it's going to change our world and how we experience it. Many of us still think of privacy as a right afforded to us that should be protected from government encroachment as much as possible (i.e. cops or military shouldn't be able to just enter your house for the hell of it and see if you are doing or possess anything illegal). But such ideas may become very rare and old-fashioned in the coming world.

November 16, 2006
Permalink ||
I've updated the quotes section.

Site Updates
November 16, 2006
Permalink ||
Archived a couple blog articles into their appropriate sections, and reorganized the science page to include a section on cosmology, astrobiology and space travel. Also updated the Funny Signs page in the fiction section.

November 16, 2006
Permalink ||
"You're the only one who knows when you're using things to protect yourself and keep your ego together and when you're opening and letting things fall apart, letting the world come as it is - working with it rather than struggling against it. You're the only one who knows."
--Pema Chodron

"We are fragmented into so many different aspects. We don't know who we really are, or what aspects of ourselves we should identify with or believe in. So many contradictory voices, dictates, and feelings fight for control over our inner lives that we find ourselves scattered everywhere, in all directions, leaving nobody at home."
--Sogyal Rinpoche

"With no consideration, no pity, no shame,
they have built walls around me, thick and high.
And now I sit here feeling hopeless.
I can't think of anything else: this fate gnaws my mind -
because I had so much to do outside.
When they were building the walls, how could I not have noticed!
But I never heard the builders, not a sound.
Imperceptibly they have closed me off from the outside world."
--Constantine Cavafy

"When we get out of the glass bottles of our ego,
And we escape like squirrels turning in the cages of our personality
and get into the forests again,
we shall shiver with cold and fright
but things will happen to us
so that we don't know ourselves.

"Cool, unlying life will rush in,
and passion will make our bodies taught with power,
we shall stamp our feet with new power
and old things will fall down,
we shall laugh, and institutions will curl up like burnt paper."
--D.H. Lawrence

"Only to a magician is the world forever fluid,
infinitely mutable and eternally new.
Only he knows the secret of change.
Only he knows truly that all things are crouched in eagerness
to become something else
and it is from this universal tension
that he draws his power.
--Peter Beagle

"Journeys bring power and love
Back into you. If you can't go somewhere,
Move in the passageways of the self.
They are like shafts of light,
Always changing, and you change
When you explore them."

[with thanks to Whiskey River for the quotes]
Poker Odds, Bad Beats, and Determinism
November 10, 2006
Permalink ||
I'm going to assume some familiarity with poker, specifically Texas Hold'em since it's the most popular right now.

Okay, so let's say you are playing heads up (one on one) against someone and you are dealt A-10. You go all-in. Your opponent calls and shows K-J. The first four cards come up 2-6-9-K, all different suits. Now, all your money is in the pot pre-flop so there's no betting to worry about here, but let's say you want to know your chances of winning this hand, given the information above.

In other words, what is the probability of getting an ace on the river? There's three left in the deck, and after subtracting out your hand, your opponent's hand and the cards already up, there's 44 cards left out there where the aces are mixed in.

Well 3 divided by 44 is .068, so you have about a 7% probability of winning the hand here, right? That's the normal story, it's how most players calculate pot-odds when they're faced with the option of calling with a hand that might not win, and it makes perfect sense. But I'm going to argue it's wrong.

The Probability Of An Ace Is Either 0 Or 1

Let's say I'm the dealer in the hand described above. I peek under at the river card. Either I see an ace there or I do not, right? Now, when the card actually flips up a second later, it hasn't magically changed to something different than what I saw -- it's still whatever it was before (either an ace or not).

So as soon as I looked, I knew the probability of you winning the hand. If I saw an ace, the probability of you winning this hand is 1 -- there's no way you can lose it. If I did not see an ace, the probability of you winning this hand is 0 -- there's no way you can win it. Nowhere does 7% enter into things; your probability of winning is not .068, sorry.

Now, from your perspective, before I flip it over, you do not know whether the card under there is an ace or not, so you can say that the probability of winning this hand is either 1 or 0, but you do not know which of those it is.

Of course, if I, the dealer, had not looked at the card, would things be any different? Surely the cards that are sitting there are whatever they are -- they don't change, they don't sit in a state of quantum indeterminacy like Schrodinger's Cat until the moment they are observed. The card about to come up either is an ace or it is not an ace regardless of whether I look; and whether it is an ace depends simply on the order of the cards before shuffling, and how they were mixed up during shuffling (there's no such thing as a truly random shuffle, just one's that approach pseudo-randomness better or worse). The aces ended up somewhere in the deck, and they sit there. Either the 12th card down in the deck (12th because: 2 to each hand, 3 on the flop, 1 on the turn, 3 burn cards, then the river makes 12) was one of the aces when the cards were mixed around or it was not.

The state of the deck is fixed and decided. It is our knowledge which is open until we have further information. In other words, what we have is metaphysical determinacy and epistemic indeterminacy. Gaining further information about the state of the deck (by showing a card) removes epistemic indeterminacy and we see what the situation was all-along. In this case, when I peeked I saw an ace, so it turns out that the entire time, even before I'd finished laying the first burn card on the table, you were guaranteed to win the hand! It wasn't a 7% long-shot win at all.

So Where Did That 7% Come From?

What the calculated number above actually tells us is: given a million hands that start off the same as the one I described (A-10 vs K-J and seeing 2-6-9-K), roughly 7% of those hands will end up having an ace as a river card (i.e. twelfth down from where the deck was shuffled). So approximately 70,000 out of 1,000,000 hands starting the same way will end with you winning.

When we calculate odds in poker, we are not saying that in this particular hand, you have a X% chance of winning. We are saying that out of the probability of space of all possible events from here on out (i.e. given the cards shown), and assuming that all of those possible events are equally likely, we know that X% of them involve an ace being the river card. Sometimes it's followed by 4-9-7-Q-4-7; sometimes it's followed by A-A-3-10-9-2; sometimes it's followed by J-3-Q-5-8-A; none of that matters to us, except that those are the possible orientations of the entire deck from which we build the probability space that gives us our 'magical percentage'.

And that's what it is, a magical percentage. Poker players and other gamblers treat it as incredibly meaningful, such that if they calculate out their own odds as 80% chance to win, they are shocked or angry when they lose. Well, if you play that situation a million times, there's 200,000 hands where you'll lose, so this was just one of those. There's no reason in a particular hand to think that it is one of the 800,000 you'll win, because the 12th card ended up whatever it ended up being long before anything was dealt out.


(1) Players need to be less bitchy about "bad beats".

(2) When faced with a situation where pot-odds and the like come into play, remember that such odds are only a sure thing if you are able to play out a near-infinite number of such hands. It may actually be in your best interest to fold a hand even when you have good pot-odds to call.

Let me explain (2) in case it is not clear. Let's say you are not rich, and you normally play poker tournaments for $5 buy-ins. But tonight you won a lucky ticket to compete in a tournament for millions of dollars. You get to a hand where you already put some chips in the pot, and so did some other players, such that when an all-in raise gets back around to you and most people have already called it, you only have to put in your last few chips for the chance to win 300 times as many chips. You look down at some meager hand like 5-6 offsuit, let's say it's a hand that wins 1 in 150 times over many, many such hands.

Pot odds scream that you have to call, because your chance to win is much better than the small ratio between the chips you have to put in and the chips you'd win. But of course, odds deal with long-run outcomes if you could play this sort of hand tons and tons of times. You won't be able to, because if you lose this hand, you lose this tournament, and the only time you'll face this sort of hand again is in a $5 buy-in tournament where, yes, playing it over and over again will pay off in the long run, but never with the amount at stake in this tournament.

If the deck in this hand happens to have been shuffled such that the cards landed in a way that will not lead to you winning, then you have just lost your one and only chance to win big, even if the odds made it rational to call. It might, in fact, be much more rational to play a hand with lower pot-odds, if it has lower pot-odds because you get a smaller payoff if you win (you get back 10 times the chips you have to put in, not 300 times) but maybe a much better chance of winning (1 in 3, not 1 in 150).

Hopefully this example illustrates why it helps to always remind yourself that an individual hand is not a microcosm of the overall probability space. In the overall probability space, your hand ends up winning in X% of them. But in this one particular hand, your hand does not end up winning X% of the time; it either wins this one time, or it does not win this one time.

Remind yourself of that, even as you take the long-term numbers into account. Because you do still have to decide whether or not to call based on something, and it is generally rational to use calculated odds if you play poker often enough for those odds to come into play. But that's just because you can afford to lose all those hands you are bound to lose (from the very start, before the flop even hits the table), by making up for it later when the same situation comes up over and over again.

Procrastinating Lowers Stress?
November 05, 2006
Permalink ||
An MIT study [PDF] a while back shows the effects of deadlines on procrastinating:

Subjects were split into three groups who had some not-so-fun task to do within 21 days. One group had three imposed deadlines, one every 7 days along the way. Another group simply had an end deadline for all the work at once, after 21 days. The last group set a self-imposed schedule of deadlines up to the 21 day mark.

The results: having one final deadline leads to the most errors and the most late submissions of the work. More importantly, setting your own schedule doesn't help as much as you might think; people still procrastinate. The best way to minimize late submission of work and do better work is to force deadlines on people along the way.

Which isn't all that surprising really, right? But a result they also showed in the study: those under the strictly-imposed deadline schedule also disliked the task the most. Deadlines work, but they piss people off (likely stress them out more).

So, I think the lesson to take from all this is that procrastination and a lack of deadlines is good, because it reduces stress to the one day before something is due. Yup, that's what I got out of the article. Which I haven't had time to read in full yet. But I'll get around to it...according to my own self-imposed deadline schedule involving a single deadline at some indefinite point in the future that moves away from the present moment at approximately the speed at which time passes.