While most people are focusing on their own individual use cases for Roam, co-founder Conor White-Sullivan has a bigger target in mind for Roam. As he writes in the Roam White Paper, “The ultimate goal is to extend the system to collaborative reasoning, allowing groups to build shared mental maps and make faster and better-informed decisions”.

In this article, Jordan Olmstead discusses how reaping the benefits of collective intelligence will require more than just the use of new tools like Roam. Also critical will be the development of new processes and institutional approaches that enable the creation of better and more effective models of reality through the embrace of challenge and uncertainty.

 

Musings on collective intelligence inspired by Rajiv Chandrasekaran’s Imperial Life in the Emerald City

Over the last few weeks, I’ve fallen down the collective intelligence (CI) rabbit hole. Companies (Roam Research) and open-source projects (Athens, Org-Roam) are promising a future where individuals can tap collective knowledge for fun and profit.

To that end, CI organizations are building tools that will allow users to represent their thoughts in graph form (a digital brain). Users will be able to traverse that digital brain for unexpected insights. The tools will also structure their thoughts in a form that can (eventually) be shared with the second brains of other users.

I’m fascinated by CI’s promise. CI will expose users to new approaches for familiar problems. CI will surface hidden connections to users.

Knowledge markets will emerge as enterprising users discover opportunities for information arbitrage and knowledge-sharing protocols enable users to encode their knowledge into digital assets.

But I have to ask – will better information necessarily lead to better decisions? Will knowledge creators be compensated fairly for their labor? How do we maintain the information integrity across a constellation of distributed knowledge graphs?

So, I’ve decided to start a blog to promote discussion on the psychological, economic, and political barriers to effective CI – and how those barriers can be addressed through thoughtful design.

This article comes from my first blog post.

Uncomfortable truths

I recently finished Imperial Life in the Emerald City by Rajiv Chandrasekaran. (The indented passages below are quotes from the book.)

Imperial Life in the Emerald City provides a first-hand account of how the United States’ Coalition Provision Authority (CPA) mismanaged Iraq’s post-invasion transition from US protectorate to self-governing state.

Chandrasekaran attentively chronicles how senior Bush administration officials and CPA staffers established norms that discouraged dissent, interdisciplinary collaboration, consultation with subject-matter experts, and genuine engagement with the Iraqi public.

Chandrasekaran, who was stationed in Iraq as a foreign correspondent for the Washington Post prior to Iraq war in 2003, was uniquely well positioned to document how the CPA’s institutional pathologies led to ill-fated decisions – most notably de-Baathification and the dissolution of the Iraqi military – that shattered Iraqi institutions and laid the seeds for insurgency.

In other words, the CPA didn’t fail to understand the situation in Iraq, the CPA did what it was designed to do – conjure a plausible image of Iraq that could be used to advance the objectives of senior CPA staffers, US officials and select Iraqi elites.

Can CI solve for people in organizations that are – at best – indifferent to reality?

Annotated passages

Don’t take their word for it

After the invasion, US Army Major Peter Veale was responsible for choosing a headquarters for the CPA.

Veale asked coalition intelligence analysts for a list of all palaces that remained standing.

While he was waiting for a response, Veale located a promising possibility:

On a website called DigitalGlobe, [Veale] pulled up images of a massive edifice with a blue dome – the Republican Palace. It seemed ideal. But when he received a list of standing palaces from the intelligence staff, that one wasn’t among the possibilities. He walked to their villa with the DigitalGlobe printout. “Hey, you didn’t show me this one,” [Veale] said. . . “Yes, this is a place and it does exist [one of the analysts said]. But it got hit on the first night of shock-and-awe [bombing] and it’s been pretty much destroyed.” (p.48)

For background, the Republican Palace was kind of a big deal. It was where Saddam “preferred to meet visiting foreign dignitaries.”

Despite Veale’s skepticism – his image of the Republican Palace was less than a week old – he chose a different location for the CPA headquarters after the intelligence team told him they were “certain” and he should “take their word for it.”

A few days later he arrived at the palace and found it destroyed.

As he sat outside the palace, a few Special Forces soldiers stopped to talk. Veale recounted his plight. One solider encouraged him to keep looking. “Hey man, you have to go down the road… there’s a palace that’s fully intact.” When Veale got there he discovered that it was the same palace he had seen on the DigitalGlobe website. (p.49)

It’s comfortable to treat expert assessments as authoritative. But even experts forget to corroborate trusted sources. We need to design CI platforms in a way that constantly exposes contradicting or complicating information instead of sweeping it under the rug.

We’ll just get someone else to write another fatwa

Ayatollah Ali Sistani was (and continues to be) one of Iraq’s most influential men. Many Iraqi Shiites looked to Sistani, who had “enormous authority in Iraq to interpret Islamic law in everyday life,” for guidance.

Iraqi Sunnis and Kurds respected Sistani because of the reverence Iraqi Shiites held for him.

The US took awhile to grasp his importance. Sistani also refused to meet directly with the Americans, forcing them to rely on go-betweens.

The viceroy’s [Jerry Bremer] first emissary was an Iraqi-American from Florida. . . he was neither a diplomat or a politician, but a wealthy urologist who had developed and patented a penile implant for impotent men. . . The CPA replaced him with a pharmaceutical executive from Michigan. Neither man, Shiite politicians told me later, projected the requisite gravitas. “If you were occupying Italy, would you send a doctor to visit the Pope?” (p.91)

Sistani later issued a fatwa (religious edict) demanding that Iraq’s constitution be written by representatives elected by the Iraqi people, and warned that a US-chosen drafting committee would be “unacceptable” because there would be “no guarantee that such a committee will draft a constitution upholding the Iraqi people’s interests and expressing their national identity and lofty social values.”

CPA staffers were unfazed.

“It didn’t register,” one of [Jerry] Bremer’s senior aides said later. ‘The view was, We’ll just get someone to write another fatwa.”

Lesson? CI will fail without mechanisms for sensitizing users to cultural context. Each culture – whether a scientific discipline, a niche Twitter community, an ethnic group – has individuals and groups whose opinions are uniquely valuable in that culture.

Users need to understand who these individuals and groups are, why they are special in that context, and the subjects for which their knowledge is particularly authoritative, lest users who are unfamiliar with a topic area discount an assessment from a uniquely authoritative voice.

I know a guy

You might have assumed that the CPA had a lot of Iraq experts, since it was, you know, trying to fix Iraq. You’d be wrong.

Most of the CPA staff had never worked outside of the United States. More than half, according to one estimate, had gotten their first passport in order to travel to Iraq.

According to Frederick Smith, the deputy director of the CPA’s Washington office:

The criterion for sending people over there was that they had to have the right political credentials.

There were few experts on Iraq, the Middle East, or post-conflict reconstruction.

Middle East expertise was actually a demerit for prospective candidates. In one memorable case, the individual chosen to head Iraq’s new police force

“lacked policing experience in post-conflict situations, but the White House viewed that as an asset. Veteran Middle East hands were regarded as insufficiently committed to the goal of democratizing the region.”

Most positions were filled by senior officials scanning their Rolodex for someone who was a “good guy” – if they were loosely knowledgeable about key job functions, that was a plus. If they were aligned with Republican party foreign policy sensibilities, even better.

CI systems must discourage users from disproportionately relying on information from individuals or organizations who are familiar to the user, but unfamiliar to the issue the user faces.

You have what you have

One of the CPA’s first priorities was privatizing companies owned by Iraqi government industries.

Privatization, the thinking went, would bolster the Iraqi economy. Furthermore, without privatization, reconstruction spending would be unsustainable.

At the same time, Iraq was in crisis. Baghdad was rocked by electricity shortages. Unemployment was rising.

The CPA assigned three men to lead the Ministry of Industry and oversee its privatization.

One of them was Glenn Corliss, a former VC associate.

Peter McPherson, Jerry Bremer’s economics czar, wanted an analysis of all 150 factories owned by the Ministry of Industry’s forty-eight companies, in order to determine which were the most viable… “This was the very first evidence that I knew that these guys had no idea what they were talking about or what they were in for,” he told me. “They said, okay, you have a hundred fifty factories. We want you to evaluate all of them.”

It gets better!

“Okay, well where’s my staff?” “It’s you.” “Okay, Where’s the management of these companies?” “Well, we fired most of the management because of de-Baathification.” “Okay. Where are the financial statements?” “There are no financials.” “Okay. What do you know about the companies?” “Nothing. We have a list of names. Here are the names.” “How much time do I have?” “You have two weeks.”

So Corliss clearly doesn’t have enough information to make a good decision.

But it gets better!

When he got a hold of the Ministry’s translated financial statements, they said each of its companies was profitable – the Ministry created the illusion of profitability by subsidizing them at the pre-war exchange rate.

Corliss found that his Iraqi intermediaries genuinely thought they were profitable because they didn’t understand how they were being subsidized by the Ministry.

After visiting the factories and crunching the numbers, Corliss realized that most of the factories were a lost cause.

Higher-ups at the CPA eventually realized that privatization, if rushed, would result in massive layoffs and social unrest. Without consulting Corliss or any Iraqi stakeholders they came up with a compromise – all state-owned enterprises would start from a “clean slate” and have their debts expunged.

The thinking was that the bad companies would be driven out of business and the good companies would survive.

However, many of the worst businesses owed money to the good businesses – when their debts were expunged, good companies suffered and bad companies benefited.

So we have a situation where no one really understood what was going on; the CPA didn’t know anything about Iraq’s SOEs (in part because they fired the people who led them), the SOEs didn’t understand that they weren’t actually profitable, the CPA didn’t understand the intricate threads of debt entangling Iraq’s state-owned enterprises, and as a result, advanced a policy that actually hurt some of Iraq’s most promising companies.

The point of all of this is we could choose to privatize or not to privatize. It didn’t matter. We didn’t have the power to do anything…They were like, ‘What are you talking about? There’s three of you. There’s 150,000 of us. You haven’t seen most of the factories. Why do you think that you’re going to make any of the decisions?’ So they just kept doing their thing, and we sort of played in our little, imaginary world over at the CPA.

CI systems must encourage users to question assessments formed with insufficient information or relying on sources with epistemic gaps on subjects related to the assessment.

Conclusion

‘I just don’t think we sent the A-team’, he said. ‘We didn’t tap – and it should have started from the White House on down – just didn’t tap the right people to do this job.’

Indeed.

It’s hard to gather the right group of people to answer a question. It’s hard to get them the information they need to answer a question. It’s hard to get people to listen to all of the people they should listen to. It’s hard for people to reconcile conflicting information.

CI system designers must reckon with a user who cares more about justifying what they want or what they believe than understanding reality.

Collective intelligence platforms will fail to improve decision-making unless they are designed to consistently expose uncomfortable information to users.

As such, effective CI will require more than an innovative mark-up language or advances in knowledge graphs:

  • CI systems should teach users how to evaluate incomplete and contradictory information through source corroboration.
  • CI systems should constantly present users with information that challenges their beliefs, preferences, and aspirations.
  • CI systems should constantly surface information from unfamiliar sources and provide users context for understanding their relevance.

 

Men at some time are masters of their fates:
The fault, dear Brutus, is not in our stars,
But in ourselves, that we are underlings.

Julius Caesar (I, ii, 140-141)

 

Image – CPA staffers in Iraq (DOD archives).