Yesterday news broke that Change.org, an historically progressive-leaning distributed organizing platform, would shift to working with any advertising client, regardless of political affiliation. The story has been ably covered by Ryan Grim, Jeff Bryant, and Aaron Krager – I highly recommend you read their pieces, all of which hinge around leaked internal Change.org documents that cover this shift.
The documents are well worth reading and have been posted by Krager (all links are PDFs). They include:
- A July 2012 email from CEO Ben Rattray to staff explaining the recent decisions by senior staff to pursue a big change in their client advertising policy;
- A September 2012 email from Rattray to their staff explaining the shift;
- An internal FAQ explaining their policy and technology changes;
- A revised advertising policy.
As I said, the posts linked above give a good run down of the general problems associated with this shift in policy and values form Change.org. I recommend you read them and the leaked documents, which give a very clear view of the goals and motives behind this shift.
I want to draw attention to one particular aspect of Change.org’s justifications for this move, quoting Jeff Bryant:
What will change is that Change.org will no longer “filter potential advertisers” based on the advertisers’ “values.” Nor will Change.org filter potential advertisers based on any “gut feelings about the content of the ad itself.”
The implication expressed in Change.org’s internal documents, by Change.org’s spokesman Ben Joffe-Walt who Ryan Grim quotes as saying, “Change.org is “not beholden to one community,” and by the talking points circulated by multiple Change.org staff members on progressive email list serves all point to the idea that it’s simply not possible for Change.org to make determinations about which clients are or are not progressive. As a result, they are saying they are now formally stopping to make any attempt to limit who they sell email addresses to based on their “values.”
These talking points are undermined by their expressed strategies for evolving their advertising platform. In a section in their internal FAQ titled, “When will we be able to target ads better?” they have this explanation:
- Machine learning: we are developing the technology to match action alerts to users, which utilizes everything we know about a user (what petitions they’ve signed, geography, demographics) to match them to petitions they’re most likely to be interested in. This is complicated technology but should bear fruit in 2013. Once that happens, we should be able to repurpose the technology and use everything we know about a user (what petitions they’ve signed, geography, demographics) to match them to the ads (sponsored petitions) they’re most likely to be interested in.
- Tagging: we want to move from our current 8-cause system to a much more flexible tagging system. Once complete, users and Change.org staff will be able to tag any petition in many different ways, for example as “pro-choice.” We will then be able to show that “pro-choice” advertisement to people who have signed petitions tagged as “pro-choice” while suppressing people who’ve signed “pro-life” petitions. This is technically complicated, and we’re hoping to make significant progress in 2013.
To be clear, what this means is not only that Change.org is saying internally that they are capable of assessing the political orientation of an advertiser or a petition, but that this assessment is something which is critical to their evolved business model.
I raise this point because to me the idea of determining what is or is not in line with the values this company espoused since its founding until this week is completely possible. It’s been done with relative success by Change.org – excepting their work with union busting clients like Students First and Stand for Children – throughout the history of the firm. And most importantly, their ability to determine if a client should target liberal or conservative audiences is central to their future business model. They will be selling organizations and companies this ability – it’s what will make their ads worth money to their clients.
When I look at Change.org’s talking points and internal messaging documents, I see a lot of sophistry and disingenuous argumentation that I’m not going to go through now. I see statements like they’re not doing this for the money and since I am not a mind reader, I can only speculate whether or not that is true.
But Change.org is telling the public that they are simply incapable of figuring out if their clients are liberal or conservative and as a result must throw up their hands to even trying to make the choice – this is a flat-out lie. Their own technology development and advertising targeting plans reveal it as a lie. Not only are they capable of making a determination as to what a client’s values are, it’s what they are selling their clients to maximize the impact they have as an advertising platform.
There’s a lot to be unhappy about with this devolution at Change.org. I’m sure others will write more about it in coming days and I’m guessing I will too. But the completely cynical use of a lie about their fundamental ability to figure out who they are partnering with when they sell ads is something that I feel compelled to highlight first and foremost.
All with all of my blogging, this post represents my views alone and not that of my employer, Citizen Engagement Lab.
Hold Fast 
thanks for the summary and links.
sounds like a big sell-out to me. i’ve signed my last change.org petition.
LikeLike
As shocking as it is shameful. I thought Change.o was better than that.
Progressives care about integrity. Integrity matters.
Soon, contrary to the assertion of Ben Joffe-Walt, Change.o will indeed be “beholden to one community”. The conservative community.
As neither group cares a whit about honesty, sincerity, or forthrightness, it’s really a perfect fit.
Hopefully they both see a future deserving of their mutual character.
LikeLike
Y’know… What really kills me is the self-serving BS rationale of Mr. Rattray. With a company history like theirs, he has the gall to claim that their goal is “to become a ubiquitous global platform that becomes a fundamental part of the infrastructure of civil society around the world, radically democratizing access to power for hundreds of millions of people”.
Specious claptrap.
In the context of the America we live in circa 2012, “democratizing” the people IS a progressive value. In instance after instance, the goal of conservative ‘movements’ is to DENY peope access to the decision making table. Conservative goals are inherrently about keeping people out, and maintaining the power structure of the elite.
What consevative ‘movement’ can Mr. Rattray point to as an example of the alleged universiality of people-powered movements?
Can he name one? The goals of his highly profitable client “students first” are anything but; they hope to take advantage of people’s ignorance, privatize/profitize the education system, and marginalize federal protections that safeguard the rights of ALL students and teachers.
What other ‘movement’ could he point to? The Society for the Advancement of Poll Taxes? The State’s Rights Lunch Counter Protection Brigade?
These ‘crusades’ are about shutting people out, and removing the barriers corporations face when attempting to monetize the entire planet.
Like his company, his prattle is transparently bogus.
LikeLike
This is one of the dumbest arguments I’ve read in response to Change.org’s new policy, and demonstrates a complete lack of understanding about machine learning.
What Change.org is saying is untenable is separating all nonprofits, organizations, and corporations into two groups: “good” and “evil”. And then making the subjective call for every single organization in the world about which group they fall into. I’d like to see anybody try to do that. Not even all so-called progressives could agree on these two groups.
If the goal of machine learning was to separate all users and petitions into two groups, then I guess you would have a point. But that’s not how machine learning works. The goal of machine learning is split users into literally thousands of un-named groupings. Groupings that humans would never be able to identify and draw a box around. Within that context petitions can be targeted to the appropriate group of users based on whatever machine-recognized characteristics they share.
It recognizes that there aren’t easily defined boxes to draw around people and their opinions. Not every progressive can be expected to hold the same views on animal rights, immigrant rights, environmental issues, LGBT rights, gun rights, etc. Unfortunately, or I would say fortunately, people can’t be categorized into two easily labeled groups. Neither can organizations. It seems like Change.org is just recognizing this and doesn’t want to be in the position of trying to force every user and organization they work with into that system.
LikeLike
You’re completely missing the point of how Change.org is planning on using technology and human tagging to sort out ideology on their site. They are clearly going to associate certain positions on one issue with certain positions on other issues – ie, someone who signs a pro-choice petition and an anti-fracking petition will be have in a similar way on those issues as well as others. That’s the whole point – without human guidance as to the associations of activities on certain issues, the predictive value of programming is worthless. There must be human guidance for how values connect to each other in broad strokes.
The notion that managing a complex matrix of relationships for ideological perspectives is less complicated that a binary set of values is really dishonest. That a machine would be able to do that better than humans without human input is simply bizarre.
LikeLike
John Eggleton’s response is sophistry.
It ignores the history of the company, the reasoning on which the company seems to have based it’s ideological switch, and the obviously relevant context of the times in which we live.
These are not esoteric musings on the fundamental nature of good and evil, these are questions of profit vs people. Change.o has made a financial decision, and that decision is inherently partisan.
Stripping away all the relevant context in order to make this an argument over the nature of machine learning is fallacious – at best (in addition to his misunderstanding as to the nature of the tech itself, as Mr. Hamlin pointed out above). Silly, even.
Additionally, contrary to the inevitable conclusion of his specious argument, it IS entirely possible to – broadly speaking – classify people and the causes they pursue into two groups. For an example of this division, pick up your nearest newspaper.
— y’know… I’m getting tired of fighting disingenuous nonsense like this. Few people seem to care about anything anyway. We’re just going to have to hit bottom, I think.
Have at it, intentional ‘mischaracterizers’, lol…. I’m out.
LikeLike