Poor Little Burckhardt

by Henry on January 3, 2013

Perhaps these heebie-jeebies are mine and mine alone, but the parallels between this Sasha Issenberg piece

The Obama campaign embedded social scientists from the Analyst Institute among its staff. Party officials knew that adding new Democratic voters to the registration rolls was a crucial element in their strategy for 2012. But already the campaign had ambitions beyond merely modifying nonparticipating citizens’ behavior through registration and mobilization. It wanted to take on the most vexing problem in politics: changing voters’ minds. … as campaigns developed deep portraits of the voters in their databases, it became possible to measure the attributes of the people who were actually moved by an experiment’s impact. … An experimental program would … develop a range of prospective messages that could be subjected to empirical testing in the real world. Experimenters would randomly assign voters to receive varied sequences of direct mail—four pieces on the same policy theme, each making a slightly different case for Obama—and then use ongoing survey calls to isolate the attributes of those whose opinions changed as a result.

and this classic Frederik Pohl short story

It was the morning of June 15th, and Guy Burckhardt woke up screaming out of a dream. It had been a monstrous and incomprehensible dream, of explosions and shadowy figures that were not men and terror beyond words. He shuddered and opened his eyes. Outside his bedroom window, a hugely amplified voice was howling. Burckhardt stumbled over to the window and stared outside. There was an out-of-season chill to the air, more like October than June; but the scent was normal enough — except for the sound-truck that squatted at curbside halfway down the block. Its speaker horns blared: “Are you a coward? Are you a fool? Are you going to let crooked politicians steal the country from you? NO! Are you going to put up with four more years of graft and crime? NO! Are you going to vote straight Federal Party all up and down the ballot? YES! You just bet you are!” Sometimes he screams, sometimes he wheedles, threatens, begs, cajoles … but his voice goes on and on through one June 15th after another.

are a little more immediate than I would like.

I’m bullish about how experimentalism can improve democratic practice, when it happens under conditions of rough power equality. But it can equally be used to improve techniques of manipulation. One of the big themes of Pohl’s 1950s science fiction (The Space Merchants, “The Wizards of Pung’s Corners”) was how unpleasant the world could become if advertising actually worked. We may be about to find out if he’s right.

{ 41 comments }

1

PGD 01.03.13 at 9:33 pm

The issue is even worse than that. I think a lot of what the Obama campaign did was network mapping of peoples’ facebook connections to try to find voters who were likely to share ideological views with active supporters but were more apathetic. Then they would use the active supporters to prod their friends/acquaintances to vote. Think of what is being produced there — a mapping of ideological networks across the society, based on private communications. Some former government person said appropos of mass government monitoring of the internet and phone connections that we were headed toward a ‘turnkey totalitarian state’. Data mining of all the words produced on the net offers some nightmarish possibilities for social control.

2

Tony Sidaway 01.03.13 at 9:37 pm

Go on, you’re daring me to say it. Okay, here goes:

Only this year’s Feckle Freezer is any good at all! You know who … Commies own Triplecold Freezers! Every freezer but a brand-new Feckle Freezer stinks!

3

Sebastian H 01.03.13 at 9:41 pm

I’m sure we can trust the politicians to write fair rules limiting their own advertising opportunities and the opportunities of their opponents to roughly the same treatment.

4

Tony Sidaway 01.03.13 at 9:42 pm

“Think of what is being produced there — a mapping of ideological networks across the society, based on private communications. “

Well, not quite. Or maybe I misunderstand you. Are you saying that what I put on Google+ or (if I still used it) Facebook, is private?

I do agree that there are concerns about privacy here, but it’s important to recognise that much of what we do now in public is electronically accessible. Perhaps you should change the law a bit to protect privacy. This has been done already in many countries, including the one I live in. But then again, we give so much of ourselves away. We are become a quite vast community.

5

shah8 01.03.13 at 9:54 pm

Wait, what?

Advertising does work. Repeatedly dunking everyone into Financial Cliff Everyone, We’re About To Go Over rhetoric does work. Just because you or I have critical faculties that work some of the time doesn’t mean that some social meme variant failed. You don’t need everyone to be a sucker. You just need enough suckers at the right time, whether that be buying overpriced cans of sugarwater or getting fools to stampede anyone who doesn’t agree with your set agenda.

6

Tamara Piety 01.03.13 at 9:58 pm

I hate to break it to you but advertising does actually work. We have for too long been (I think) relatively complacent about the sorts of pernicious effects it might have on the society, the political culture, our health and our self esteem by focusing on the observation that it doesn’t always work, all the time, on all people. But it works often enough for marketers to play annoying music to get you to buy more, to put speed bumps into reward redemption processes so you will give up before you redeem your rewards or collect your rebate, to endlessly flatter, cajole, frighten, seduce, and badger you in a thousand ways. Now the repertoire includes data mining so the data miner can use what you say in more unguarded, non-commercial interactions, to guide commercial interactions. It is no accident that politics looks a lot like commerce these days. But with the Sorrell v. IMS Health decision last year the Supreme Court has announced that this sort of spying on you in order to sell you stuff is not only A-okay, it is constitutionally protected. Apparently being spied on for commercial purposes is a “necessary cost of freedom.” It remains to be seen whether (with the appropriate challenge) the Court will remain so sanguine about governmental or political party uses of data mining like this. It may depend more on the politics of the occupant of the White House and the members of the Court than a principled distinction between the two cases.

7

Bloix 01.03.13 at 9:59 pm

Actually what the article is about is the precise opposite of the situation described in the Pohl story. The Pohl story is about the crudest forms of mass advertising techniques that blanket a population. The Issenberg piece is about highly discerning tactics that use analytics to identify individual voters who are susceptible to persuasion.

The thing about mass advertising is that it’s incredibly irritating to the 99% of people who are not about to be persuaded. Individual targeting is much less irritating – often people who are persuadable don’t mind being contacted. If you did any canvassing for Obama – which was all about contacting likely supporters who needed a little encouragement to get to the polls – you’ll know that the canvassees often liked being contacted – they were happy to know that Obama had a well-organized campaign and that it cared about people like them.

8

William Timberman 01.03.13 at 10:06 pm

Orwell again (and again and again). Control the vocabulary people think with, and you control how — and what — they think. Mind you, you don’t necessarily control how they act, but that’s what surveillance cameras and SWAT teams are for — politics recast as a game of administrative whack-a-mole. Annoying, but no danger to the state as presently constituted. (Unless, of course, you’re dealing with the Taliban.)

9

Alex 01.03.13 at 10:07 pm

That said, mobilising the economic majority through field organising > demobilising the public through Fox News. Also, I suspect politics being influenced by experts rather than pseudoexperts-as-ideological-gatekeepers is probably better.

This is interesting and something I didn’t know:

To that end, he hired the Analyst Institute, a Washington-based consortium founded under the AFL-CIO’s leadership in 2006 to coördinate field research projects across the electioneering left and distribute the findings among allies.

The Pohl short is good. See also J.G. Ballard’s “The Subliminal Man”. “The signs, Doctor! Have you see the signs?” I thought of using that as the strapline for my blog.

10

Alex 01.03.13 at 10:08 pm

A question: does using the surname of a frequent commenter who may share it with a former head of the Imperial German Navy trip a moderation script?

11

Alex 01.03.13 at 10:08 pm

An answer: yes.

12

jack lecou 01.03.13 at 10:19 pm

Well, not quite. Or maybe I misunderstand you. Are you saying that what I put on Google+ or (if I still used it) Facebook, is private?

With services like Facebook, the rules describing the disposition of your data are rather poorly specified and ever shifting. There’s, e.g.: (1) stuff a user intends to post publicly, (2) stuff Facebook makes available publicly whether you realize it or not and especially (3) info about yourself that you’re creating in the course of using the service and implicitly authorizing Facebook to package up and share with third parties for a fee.

I don’t know the details, but I expect a campaign like PGD described would have to rely a lot on (3), and include data that isn’t necessarily ‘publicly’ available (at least not in the sense that you could scrape it out of Facebook’s servers on your own, without their explicit help.)

Even if it didn’t, it’s of course not hard to imagine that Facebook or someone else won’t be selling such data soon.

13

Chris Mealy 01.03.13 at 10:30 pm

If anybody has a better idea for building a mass membership organization I’d like to hear it.

14

PatrickinIowa 01.03.13 at 10:40 pm

I’d be interested in seeing whether the age of the commenter correlates in a meaningful way with the degree of outrage at using Facebook to encourage less motivated voters to get to the polls. (I’m 59. It makes me queasy, but not crazy.)

15

JSE 01.04.13 at 12:42 am

We may be about to find out if he’s right.

As others have mentioned, we know that advertising already works, to a certain extent. The question is whether social data enables advertising to work to a much greater extent, or even in a qualitatively different way. That might be the case, but it’s not obviously the case. With some tasks, like estimating parameters of a distribution, you do better and better the more data you have. With others, like weather prediction, there seem to be hard limits: no matter how much information you gather, you won’t know the weather a month from now. For all we know, political persuasion might be like that. Groups of people are complicated.

16

bianca steele 01.04.13 at 1:03 am

Heebie-jeebie-wise, the parallels between this . . . and this . . . are a little more immediate than I would like.

There are two possible answers: (1) You’re not getting enough sleep. You have a tough job. Let your wife watch the kids. (2) That isn’t surprising, as there are only three possible plots, so you’re bound to notice some coincidences like that.

If advertising is so sophisticated, why am I stalked by ads for things I just bought and things I decided were way too expensive?

17

JW Mason 01.04.13 at 1:31 am

Does anyone know if this story was the inspiration for Dark City? If so, funny they changed the motivation from testing ads to learning the Mysteries of the Human Soul.

Re the Obama story, I don’t think it tells us about much except how these seasons’s campaign consultants are marketing themselves. Perhaps because I’ve had the dubious privilege of working with some online communications hotshots, I’m not really worried about them taking over my brain. Look out for your budget, tho.

18

JW Mason 01.04.13 at 1:42 am

Also, for an updated version of the Pohl story, check out The Cookie Monster by Vernor Vinge.

19

JW Mason 01.04.13 at 1:43 am

(huh, wonder why my link to Vinge’s “The Cookie Monster” triggered moderation. Another sign of how far we are from strong AI, I guess…)

20

Dave 01.04.13 at 1:53 am

Persuasion is creepy.

21

bianca steele 01.04.13 at 2:20 am

JW,
It does sound like Dark City, until the end (was it made into a Twilight Zone episode?), and that movie did have a kind of surprising ending, but it had three screenwriters. It’s tough to know what came from each of them. One was a director, it was his first screenplay. The other was a Brit with a lot more mainstreamy screenwriting experience who, according to the interviews on the DVD extras, made the movie releasable by recognizing it was an existential drama, and rewriting it accordingly. The third seemed at the time, IIRC, a comics or videogame oriented guy. (He later worked with the Nolans on the Batman movies.) Not terrific evidence in support of auteur theory, on that evidence, anyway.

22

bianca steele 01.04.13 at 2:26 am

The fingers at the end are a nice touch, because somewhere the image of a four-dimensional person or god reaching through our space is described or depicted with the fingers manifesting as spheres.

23

Henry 01.04.13 at 2:39 am

(huh, wonder why my link to Vinge’s “The Cookie Monster” triggered moderation. Another sign of how far we are from strong AI, I guess…)

The opposite. What kind of entities do you think we have running our comment filters? And do you really think we want to provide them with any information that might alert them to the true conditions of their existence?

24

js. 01.04.13 at 3:07 am

Experimenters would randomly assign voters to receive varied sequences of direct mail—four pieces on the same policy theme, each making a slightly different case for Obama—and then use ongoing survey calls to isolate the attributes of those whose opinions changed as a result.

From what I understand, this sort of thing—at least the general structure of testing various slightly different appeals—is pretty common in direct mail campaigns (for fundraising appeals for NGOs say), and has been for some time (though I don’t know enough to really give a time frame). So I’m not sure this is really all that new or innovative. (And are we sure “experimenters” isn’t a bit of advertising itself?)

(Haven’t read all the comments, so apologies if this point has already been made.)

25

js. 01.04.13 at 3:15 am

To follow up on my last comment:

There might of course be a very legitimate worry about these techniques being used in an explicitly political arena to directly promote a political outcome. But that seems a different worry than that “advertising might finally work”. (I have no idea if it presently does, but certainly bit of a waste if it doesn’t.)

26

JW Mason 01.04.13 at 3:21 am

I’m relieved by your reply, Bianca. Soon as I hit “submit,” I was sure the next comment would be a lmgtfy with a link to IMDB. Learning there’s no direct connection is a relief, lets me continue feeling good about seeing an indirect one.

27

Omega Centauri 01.04.13 at 3:52 am

In the spirit of js @24, I don’t think these are new techniques, just incremental improvements to old ones, enabled by advancing technology. Accumulation of sufficient incremental improvements itself can be revolutionary. I make a career out of doing that within the computer industry, which itself is a giant example of the accumulation of lots of mostly incremental improvements totally to something revolutionary. So perhaps this sort of activity crosses some threshold, where its effectiveness moves from too little to have any real effect, to becoming the main determinate of the outcome.

28

mek 01.04.13 at 5:31 am

@24, the difference for the Obama campaign was one of scale. Simply put, they can a/b test over tens of thousands or even millions, and they can do it over and over again.

CRMs have evolved by leaps and bounds in the last few years, as well. Now the campaign can model literally everyone it wants to, and not encounter technical problems because their database is 100 million people large.

The demographic targeting allowed by social media advertising is of a different sort from anything that came before.

29

faustusnotes 01.04.13 at 6:51 am

I think this is another case of panic about the medium rather than the message. Targeted advertising is much better than blanket advertising, and it’s always been the job of campaigns to change minds – that they can assess their effectiveness “scientifically” is simply an improvement in technique.

I also don’t believe they can be “scientific” about anything. They can’t control selection bias or non-response bias, and they don’t have any way to see if changes in knowledge and attitude manifested at the polling booth (and anyone who works in health promotion knows that showing a change in attitudes is essentially meaningless). Furthermore, machine learning is not that good and these kinds of predictive models are notoriously poor at prediction. Good luck to the kids at this insight company if they can spin their findings into successful funding pitches to the Obama campaign, and good luck to Obama if he can do better from it than Romney did, but it’s not quite the stuff of sci-fi yet.

(I saw articles about how Romney had the most advanced database wranglers EVAH and his technological edge was unheard of in a conservative campaign. The same techy people on his side were spinning the same tall stories…)

30

js. 01.04.13 at 7:05 am

@28: That makes sense; and is helpful. And I didn’t mean my #24 to sound as sanguine as it probably did. Still I’m curious:

Simply put, they can a/b test over tens of thousands or even millions, and they can do it over and over again.

This makes it seem like it’s basically a matter of scale (as you mention) and resources. I’m more curious about whether the real problem—or maybe: the real problem Henry wants to highlight—is about the very possibility of this sort of large scale and sophisticated targeted advertising or whether its more about the particular political application of it. (I can see different sorts of reasonds to be worried about both, but importantly they are different sorts of reasons.)

You could make the case that only a political campaign could combine the scale and resources necessary for this sort of thing, but I don’t really see why that need be true.

31

js. 01.04.13 at 7:06 am

Total blockquote fail. Second paragraph is quoted from mek @28.

32

John Kozak 01.04.13 at 1:58 pm

“The Tunnel under the World” was adapted for TV as part of the “Out of the Unknown” series in the 1960s. Used to be on youtube, doesn’t seem to be any more, but is probably out there somewhere.

33

Andrew Burday 01.04.13 at 5:49 pm

Henry, like some other commenters I share your heebie-jeebies but am not sure why they were particularly provoked by this piece. Political propaganda has been around for a long time and it’s been reasonably effective for a long time. On the other hand, the recently well-documented use of sophisticated experimental marketing techniques in ordinary commercial advertising ought to concern anyone who thinks that free markets will maximize the general happiness. Privacy isn’t only a matter of knowing who you like to have sex with or who you’re going to vote for; if we take markets at all seriously, it’s also a matter of knowing how much you’re willing to pay for toothpaste. I personally don’t have much faith in markets, but the point remains that in the system we’re supposed to be living under, violations of commercial privacy deserve to be taken seriously. The Obama campaign’s practices combine these two concerns, but they ought to have been concerns anyway (and we ought to have expected them to be combined).

JWM, 17, I like a lot of your contributions but in this case one of us is missing the point. Political marketers don’t want to take over the brains of people who are determined to resist them. They want to take over enough other brains so that in a two-party system (or any other practically workable system of government), the opinions of those who resist takeover become irrelevant. They have been successful at this in the past and marketing experiments probably will make them even more successful at it in the future.

34

C.L. Ball 01.05.13 at 12:14 am

I know one Times reporter or someone else has a book out on the micro-targeting of voters, but I’m skeptical that the practice is as accurate or as effective as portrayed by many. These databases must have a high noise level, and some of the effects are not surprising:

The Obama team found that voters between 45 and 65 were more likely to change their views about the candidates after hearing Obama’s Medicare arguments than those over 65, who were currently eligible for the program.

Well, of course, since those voters would be the ones affected by Ryan’s proposed changes. Those over 65 would continue under the old plan. That was how the Medicare “reformers” were trying to neutralize opposition of AARP.

I’m puzzled by this part:

Analysts identified their attributes and made them the core of a persuasion model that predicted, on a scale of 0 to 10, the likelihood that a voter could be pulled in Obama’s direction after a single volunteer interaction.

How did they test the model? And how would you know, for sure, whether the voter was actually pulled? Overall, though, the Issenberg piece is very good; it avoids the hype about these techniques.

One of the more disturbing aspects is the degree to which volunteering for a campaign has become, for the most part, nagging and surveilling your alleged supporters. Persuading from a script doesn’t sound like anything approaching democratic politics that we hope to have.

35

Mark 01.05.13 at 1:06 am

Except that “nagging and surveilling your alleged supporters” has always been a crucial part of campaigns. Isn’t persuasion by one-to-one contact, even if jumpstarted by a script, would be preferable to one-to-many tv advertising? (This point, admittedly, does not address the privacy concerns raised in the OP).

36

Witt 01.05.13 at 2:30 am

I’m with Andrew and others in sharing the unease but not necessarily around this specific article.

Isn’t persuasion by one-to-one contact, even if jumpstarted by a script, would be preferable to one-to-many tv advertising?

Sure, if advertising were the only variable we cared about.

But if you care about trusting that your individual interactions with other human beings are based on some degree of authenticity then it is a very big loss indeed.

I’ve been online since 1991; I admit to really not liking the way I now reflexively doubt product mentions in blog posts or Twitter streams. Constantly wondering whether a recommendation is genuine is not how I like to live my life, but I’ve gotten more and more gun-shy based on real-life experiences…so.

37

PJW 01.05.13 at 3:33 am

I watched Jim Messina, Obama’s re-election campaign manager, discuss these strategies in detail a few weeks ago on C-SPAN. At about the 11-minute mark of the Q & A program, I believe he talks about the various testing they did on their mailings: http://www.c-span.org/Events/Obamas-Campaign-Manager-Reflects-on-the-2012-Election/10737435977-1/

38

Alex 01.05.13 at 10:56 pm

if you care about trusting that your individual interactions with other human beings are based on some degree of authenticity

I do not, in fact, expect that my interactions with political canvassers will be based on authenticity.

I do, however, hope that political parties that claim to represent the economic majority will choose competence rather than authenticity.

39

Witt 01.06.13 at 2:47 am

I do not, in fact, expect that my interactions with political canvassers will be based on authenticity.

But that’s just it — this isn’t simply about someone who knocks on your door wearing a campaign button and says they’re canvassing for Obama (or whoever). It’s also about people reaching out to their own existing networks of friends or acquaintances (see PGD’s comment 1, for example).

They may not be working from a script; they may not even see themselves as campaigning in any traditional sense of the word. But they’re definitely approaching the interaction differently than, say, a generic voting cheerleader like me who can be counted upon to send a mass “Please vote!” e-mail to her friends at every election.

You can’t capitalize on people’s social bonds and trusting relationships without inevitably altering the very relationships you value — often in unpredictable ways. This phenomenon is certainly not new to the Internet or social marketing of campaigns — but it is possible to achieve in a far different way than, say, 10 years ago.

40

Hob 01.06.13 at 7:42 pm

It’s also about people reaching out to their own existing networks of friends or acquaintances (see PGD’s comment 1, for example). …. they’re definitely approaching the interaction differently than, say, a generic voting cheerleader like me …. You can’t capitalize on people’s social bonds and trusting relationships without inevitably altering the very relationships you value

Okay, I saw PGD’s comment 1. PGD said this:

I think a lot of what the Obama campaign did was network mapping of peoples’ facebook connections to try to find voters who were likely to share ideological views with active supporters but were more apathetic. Then they would use the active supporters to prod their friends/acquaintances to vote.

Leaving aside that PGD provided no evidence for this, other than “I think”— at least, I don’t see any such thing in the linked article that we’re discussing— it makes no sense.
The idea is, I guess, that I am a committed supporter of the Obama campaign, the kind of person who will spend time talking individually to my friends and trying to get them to vote— i.e., I am not “a generic voting cheerleader” and anyone who knows me already knows this— but I am unable to figure out which of my friends are more apathetic than me, until the campaign does some Facebook data analysis and tells me the answer is Witt. So then I “prod” you— by saying things like “you really ought to vote”— but, apparently, I’m trying to fool you into thinking that that’s just my own idea that I came up with myself, and that I’m not actually a campaign volunteer. In that case, it seems to me that 1. the chance of this altering our “trusting relationship” is slim since we can’t have had one in the first place, 2. the campaign has wasted some time trying to get volunteers to do what they already do, and 3. neither of us is very smart.

41

John David Galt 01.07.13 at 3:01 am

Sure advertising “really works.” The catch is that the kinds which work the best are labeled as “news” or “entertainment” or something else besides advertising. Letting the marks know that your purpose is to persuade is counterproductive (from a perspective in which that is your only goal).

It seems to me the most relevant fiction book here is one that hasn’t yet been mentioned here: Interface.

@Sebastian H: I hope you’re being facetious. Every “anti-corruption” bill since Watergate, but in particular McCain-Feingold, is really an incumbent protection act. Getting the real corruption out of the system is simple — ban lobbying, or at least fund-raising for future elections by anyone currently in office — but very hard to enact because of raw self-interest. Honest people don’t tend to win office — and the few that do don’t tend to stay honest, and the few that do both don’t tend to win reelection.

@Alex: Has the Analyst Institute made its methods public? Because it sounds to me like the good guys had better adopt them too, and fast, before our country is ruined.

Comments on this entry are closed.