Tuesday, January 22, 2019

Strategy Integration


When I began my visit to the Strategy Integration group, I expected to see something like a space mission control room. What I found instead was two floors of offices almost indistinguishable from those used by industry for management and engineering, except that the cubicles had full-size walls that blocked sound from within. A third floor was filled with conference rooms surrounding an open bay used for large group meetings.

“The data and logic behind the operation is mostly in computers,” explained Samantha Lazlo, the Austrian ambassador to WICO, head of the group’s Overview team, and my guide for three days. “Our advanced artificial intelligence system, which we call Sanda, has access to the entire WICO database, including the observations and models from the biosphere assessment, and is about to be connected to all unrestricted global online networks. 

“There are over 700 people in Strategy Integration. Most of them interact directly with Sanda from satellite offices around the world, asking questions and helping answer others that contribute to the integration and testing of the overall strategy. They also coordinate input from the national strategy groups in their regions. Our Administration team on the first floor manages who works where, handles contracts and issues, and coordinates with the other groups. And we have a technical team in the basement, which handles facilities and IT.” 

Lazlo and a member of the technical team set me up in a guest cubicle with the Overview team on the second floor with basic computer access that allowed me to communicate with Sanda, whose voice was eerily similar to Lazlo’s but without her cosmopolitan accent. One of the three large monitors occupying my desk was dedicated to visual feedback from Sanda that included text and images, while the other two functioned as a typical computer desktop with the usual applications. “Ask Sanda whatever you want,” Lazlo told me. “I’ll try to answer anything that it can’t or won’t.”

I had brought a list of questions of my own and submitted by colleagues, which I spent most of the first afternoon trying to get Sanda to answer. Virtually every one led down a proverbial rabbit hole of clarifications, details, and intersecting concepts that added exponentially more confusion to what I was trying to learn. 

Lazlo looked at me with sympathy when I shared my exasperation. “Sanda was designed to emulate in each answer the maximum complexity implicit in each question.” For some reason, my expression made her laugh. “It can’t yet assess what you already know, or what you think you know, so it assumes you know the least amount possible that would generate the question you asked. That way, you have the best chance of getting the answer you seek.” She suggested that I join her in a meeting on the third floor, where she would make formal introductions and I could get some insight from the 23 members who were currently in the building. 

The subject of the meeting was an update on the amount and quality of strategy inputs so far received by STRIDE. After Lazlo introduced me, data expert Zhou Li Xiu took a deep dive into what was being learned from the information gathered by the online tool deployed to assist development by the teams appointed by their governments. She concluded with a 45% assessment of quality based on the 56% of inputs received so far, which everyone agreed was dangerously low.

I knew enough about programming to gather that Sanda was the “back end” of the software, and guessed that it was experiencing the reverse of the problem I experienced with it: essentially, it couldn’t derive from the tool’s simple inputs the detail it needed to construct a meaningful strategy. Lazlo explained that the inputs were expected to be mostly corrections to a strategy being developed through a collaboration between Sanda and the larger group that was well-trained to work with it.

On my second day, I investigated the basis of the famous Hope Chart by datamining a computer server devoted to its creation and maintenance and discussing it with Lazlo’s team members. Hundreds of potential strategies had been simulated and each result translated into a chart that served as a touchstone for evaluation. Zhou explained that the simulations focused on interactions between “subunits” of populations and environments that were each represented by a chart and both evaluated and reality-tested by at least one pair of STRIDE group members doing field research as required.

“Our biggest concern is how natural systems are responding to what we do,” Zhou told me. “Sanda uses the latest data and physical models to assess the range of possibilities, which so far track with some simple approximations of overall ecological impact when averaged over decade-long time scales. We are seeing signs that those approximations could break down soon, which is in part why the world must act now so we can exercise what control is still possible.” 

I was reminded of the Secretary General’s analogy of heading through a minefield toward a cliff. Zhou found the analogy apt, but with a modification. “Some of those mines are actually time bombs. Familiar examples are the melting of ice and frozen methane, and of course the acidification and heating of the oceans. Some of the bombs are already exploding, as we see in the cascading effects of species die-offs, especially at the bottom of the food chain in the oceans and on land. The healing effects we hope for when we draw down our impact may be a chimera if the species we save are unable to save themselves or us.”

The third and last day was spent reviewing documents summarizing the draft strategy, and following up with Lazlo on what I had learned. My discussion with Zhou helped immensely in making sense of the strategy, even with all its placeholders. I tested my understanding by sharing it with Lazlo: “You’re basically stopping the damage being done; repairing the damage that already exists; neutralizing known and potential threats where possible; and enlisting allies to create new and safer environments wherever it can be done.”

“That’s right,” she agreed. “I would add that there’s a lot of ongoing intelligence gathering and sharing so time and resources can be used efficiently. While we presently have the luxury of artificial intelligence to help plan and drive our actions, it shouldn’t be taken as a given. People need to learn to take its place, and that will require a whole new kind of education that to me is the most critical placeholder to be filled.”

Reality Check

Organizations and technologies represented here are pure fabrication based on what I guess might be the minimum requirements for such an effort given its time constraints. 

The number of people in Strategy Integration is based on the sample size of my simulations, with overhead for other functions. It is an estimate of that it would take to gather sufficient information for realistic guidance in developing a detailed strategy. My simulations do NOT represent the culmination of such an effort; they are rather the source of approximations like those mentioned by Zhou.

Characterizations of the strategy and the issues it addresses are based on my own understanding.

No comments:

Post a Comment