Cognitive Government, Artificial Intelligence and the Future of Collaboration

Friday, October 9, 2015
by Nick Charney RSS / cpsrenewalFacebook / cpsrenewalLinkedIn / Nick Charneytwitter / nickcharneygovloop / nickcharneyGoogle+ / nickcharney

"By 2020, the cognitive technologies–machine learning, natural language processing, speech recognition, and robotics–start to augment the government workforce and improve the quality and efficiency of government systems. These technologies bring forth a range of applications in government including law enforcement, transportation, healthcare, and even fraud detection. More importantly humans remain “in the loop” not only to develop, customize, and train the systems; but also oversee, guide, and improve them." - Gov2020 Deloitte
Cognitive government (as described above) is one of the mega shifts identified in Deloitte's Gov2020 research project. Its something that Bill Eggers and Paul MacMillan elaborated on for Nesta's blog back in May, arguing that cognitive government requires open functionality, applied learning and adaptive-rule making and asking:
"How can governments shorten their learning curve to more effectively adapt to the technological changes that surround them?"*
Eggers and MacMillan seem to think that advancements in Artificial Intelligence (AI), Cognitive Systems and Machine-to-Machine learning holds the key to achieving fundamental changes to the architecture of government; and the more I think about it the more I tend to agree.

Enter Artificial Intelligence (AI)

To date, we've largely thought about our information systems as static repositories of information rather than a dynamic AI. This type of thinking has informed procurement decisions, operational decisions, and shaped the very flow of information in and thus knowledge contained within our organizations. But what if our records and document management systems were built around a dynamic AI rather than a static repository?

Cognitive computing is supposed to accelerate and enhance human expertise; capture the expertise of top performers; improve decision making; and scale easily alongside both the information supply and demand. Our current information systems do none of these things, instead they rely on the user to classify, query, judge the validity of information and bring it to bear.

What if we had cognitive systems that are able to put the entire corpus of a department's information base into context and provide confidence-weighted responses, supporting evidence and map any related actors, polices, regulations, and services?

What if the system continued to learn as new information was inputted? If it could gauge and determine the validity of a given information resource based on who produced it, who was consulted, how quickly or far it moved through the system, and whether it was ultimately approved or set aside. What if we took a more enterprise wide approach and linked departmental AIs together?

Imagining the future of collaboration

What if – to take a real world example – we had approached the web renewal initiative with an AI (rather than Content Management System) in mind?

Imagine all of the government's information in a single window that learns not only from how the civil service organizes its information, policies, services, and regulations but also how citizens search for, consume and interact with them online. The AI could automatically prioritize related information for users as they search, highlight seasonal information as it is a priority (e.g. tax filing, voting, etc), and remove redundant, outdated and trivial content (ROT).

This would take co-creation and collaboration to a whole new level as the AI brings governments and citizens together in a way that is otherwise impossible to achieve. We may not think of this as co-creation or collaboration in the contemporary sense of the word but it surely entails at least some small but important part of the future of both.

*Caveat: Melissa picked up on this point in her wildly popular, "A government that learns by design"; it is worth reading if you have not yet done so.

What Innovation Feels Like (Part 2: Lack of Trust)

Friday, October 2, 2015

by Melissa Tullio RSS / cpsrenewalFacebook / cpsrenewaltwitter / creativegov

Public servants go through a rigorous process to become full-time, permanent employees. I won't get into detail, but it usually requires a lengthy HR process with at least two to three stages of vetting to find the right person for the job. And then, if you’re one of The Chosen, you’re on the inside. Yay, you; your knowledge and expertise were tested against dozens, sometimes hundreds, of other candidates, and now that you're inside, you can get to work and start making use of all that creativity, energy, and expertise you've been specifically chosen to provide. Right?

The experience on the inside is very different than what you expect. When you try to provide that expertise, you’re beaten down by process, by politics (big-P), or by conflicting personalities (little-p). You thought you had something to offer the world, to make things a little bit better for people you’re serving, but by the time you’re able to get something you create through the processes, it doesn’t at all resemble the passion and energy you put into it.

When good people are crushed by process

Most of what we do as public servants resembles the following.
Step 1: Create this [product]. Use your knowledge, expertise, evidence, research, and best advice to create it.

Step 2: Next, slap this template approvals form on top, which requires (minimum) 10 signatures to be approved, and get every single pair of eyeballs listed on that sheet to give you their two cents on it.

Step 3: Use outdated linear proprietary approvals mechanisms to track and incorporate remarks, changes, vague comments, and illegible handwritten markups from all parties involved. Update. Revise. Update. Revise. Co-ordinate/negotiate changes among parties.

Step 4: Lose track of version (due to outdated linear proprietary approvals mechanisms). Backtrack to find original version buried in an email. Re-incorporate all changes you just lost. Repeat.

Step 5: (Miraculously and with much lost sleep) Meet deadline, feeling mangled. Start at the beginning for the next one.
Don't get me wrong; the product we start off with isn't ever perfect, and none of us should be immune to critical feedback, or fresh eyes to take a look at something to make it better. But if you step back and look at the baked-in approvals process, underlying it is an inherent lack of trust (perceived or real, which is basically the same thing to the experts that are hired to provide their best advice).

My Big Idea

Earlier this year, I pitched an idea to change the way we do approvals. It's an interesting play off of one of Kent's posts from the summer – see: Government, Citizens, and Power. What I proposed (without realizing it) would be something like applying the IAP2 spectrum internally to our processes by empowering staff to have a real stake in outcomes and decision making in government.

This approach would start with cross-functional/cross-ministry teams (5-7 max) of senior advisors – those responsible for developing products/policies at the ground level. Some authority would be relinquished from decision makers so that teams are expected to reach all final decisions on the products/policies they developed (ie, ditch the approvals sheet/mechanism all together).

The role of deputy ministers, assistant deputy ministers, and directors would shift: they'd be expected to provide feedback and direction as an advisory committee (which would be decided on before the project begins, meaning leaders from different ministries might be assigned to certain projects, depending on the nature of the work). This advisory body would check in periodically with the project team, and the feedback they provide would be more holistic, bringing in perspectives from all ministries/areas identified as "leads" on the project.

Managers/supervisors would be included on teams as quality control agents, and to ensure advisors have all the information and context they need for any piece of the products/policies. They would also play a facilitating role, determining who teams might need to work with to do their work.

An early sketch of the project flow looked like this (I'd revise this and move DMO (deputy minister's office) and legal to the advisory committee):

I'd also add to this: project teams and advisory committees would include experts (lived experience counts) from outside of government, too.

There's more to explain about how I see this proposed process working in real life (hint: we'd have to change how we do our work, too), but I'll leave it for another day and another blog post.

Let It Go

(#sorrynotsorry for the earworm.)

More and more of our work in government these days depends on all kinds of collaboration: between ministries/departments, and beyond our own level (municipal, provincial, federal) or sector (private, non-profit, community-based/grassroots). The approvals system I described at the beginning of this post does not work any more. It is seriously broken. And what needs to replace it would depend entirely on trust.

Trust means letting go of control, and the current model favours (and rewards) control freaks and micro-management.

Trust means letting go of authority, and the current model presumes that traditional power structures have the final say in everything.

Trust means letting go of fear, putting risk in its place (ie, risk management rather than risk aversion, as a rule), and experimenting with stakeholders instead of believing we have all the answers ourselves.

Richard Pietro, self-proclaimed open government fanboy and all around amazing human being, recently produced the world's first short film on open government, open data, and open source. I'm going to spoil it for you (you should watch the whole thing) and skip to the end. This is what trust feels like.
"You have to let it all go; fear, doubt, disbelief... open your mind."
...and open the processes, starting with the premise that we trust the people we so painstakingly selected for the jobs we've asked them to do.

The Ship's Doctor, Or, the Difference Between Skill and Expertise

Tuesday, September 29, 2015
by Kent Aitken RSS / cpsrenewalFacebook / cpsrenewalLinkedIn / Kent Aitkentwitter / kentdaitkengovloop / KentAitken

This is a quick reflection on HR and the difference between skill and expertise, but let's start with Firefly.

The show Firefly followed a crew of spaceship smugglers, each with a distinct role on the ship: Malcolm, the Captain; Zoe, the first mate; Hoban, the pilot; Inara, the "diplomat;" Jayne, the mercenary; Kaylee, the mechanic; and Simon, the ship's doctor.

The thing is, they only have so much room on board, and they only make so much money. And every additional crew member reduces the percentage of profits for the others - it's expensive to keep people on.

Therein lies the problem with the setup. Everyone pulls their weight, but it's a pretty crude division of skills: the only options for medical care, for instance, are to have no doctor or a full-time doctor as part of the crew and cut of earnings. In the middle of space, there's no scaling to needs. Everyone else could learn some field dressing and first aid, but it's a far cry from a trained, professional surgeon when you really need one.

The analogous situation for an organization's HR is the question of whether you find specialists or train generalists in new skills. The example that came up recently was about facilitation, and a colleague and I discussed the difference between a moderator - someone who can keep a conversation moving productively - and a trained facilitator, who can design a productive session and get the absolute most out of the brainpower in a room. But there are many such fields: data analytics, engagement, usability, research techniques, and on (or anything that fits the model described in the post Innovation and Rigour).

The problem is that very few teams, staffed permanently with a particular crew with a particular set of skills, need a full-time facilitator. But they also don't need everyone trained lightly in facilitation - it's like the difference between first aid and surgery. What these teams need is reliable, scaleable access to a surgeon when someone gets badly injured. Or, less darkly: reliable, scaleable access to specialized, expert skills on demand. For example, one would only bring in a facilitator when they host major meetings or workshops, or during strategic planning exercises.

The alternatives are dicey. To a point, a culture of collaboration solves the dilemma. But that requires that enough teams have these mutually dependent features:
  1. They've identified a need for specialized expertise
  2. They're willing to fill a valuable, rare open position with that particular expertise
  3. They have enough flexibility to lend that person, on demand, to other teams
Which is unrealistic at worst, and unreliable at best.

Otherwise, there are several models proposed to solve this dilemma: Govcloud or STRATUS, both of which require a bank of expert employees who are not permanently bound to any particular program or set of deliverables. It's comparable to the "corporate services" approach (HR, finance, IT, procurement) but for non-standardized work, where expertise meets policy and program knowledge.

Regardless, the status quo has a gap that most organizations currently cannot appropriately address. But it's the world we inhabit: incredibly high standards for program and service delivery, with increasingly niche approaches to getting there, that may not fall neatly into 40-hour-per-week job descriptions. Have people squared this circle? Are there any other approaches that are working well?

On the Rise of Policy Instrument Constituencies

Friday, September 25, 2015
by Nick Charney RSS / cpsrenewalFacebook / cpsrenewalLinkedIn / Nick Charneytwitter / nickcharneygovloop / nickcharneyGoogle+ / nickcharney

I read an interesting paper on policy innovation a couple of weeks ago entitled: Instrument constituencies and the supply side of policy innovation: the social life of emissions trading. I wanted to share it specifically in the context of my recent exploration of the #w2p online community (See: The Gentrification of #w2p) because I think it has some explanatory (and predictive) power.

Here's the somewhat academic abstract:
We offer a perspective on the making of policy instruments over time. This sheds light on the work that goes into articulating and maintaining instruments as both models and implemented policies, and the social formations that arise therefrom. Drawing on a brief case study of the innovation of emissions trading, we show the role of both functional promises to deliver public-policy outcomes and structural promises concerning new positions for the actors involved. We show how the making of instruments can coincide with the formation of ‘instrument constituencies’, which consist of entangled practices that cultivate an instrument. Constituencies sustain the instrument and are themselves sustained by the instrument as it persists and expands its realm of validity. We conclude that policy instruments can develop social lives of their own with dynamics that should be taken into account by scholars of innovation in governance.

Plain English tl;dr:
A policy constituency forms alongside the emergence of a new policy instrument. This group of actors has an inherent interest in the successful development and mainstreaming a new policy instrument because that mainstreaming creates an environment where those with instrument expertise (the constituency) are in higher demand and given positions of authority and influence; this is one of the defining characteristics of supply side innovation.

First, this isn't a judgement call on anyone

I think it's important to note that the research is value neutral in terms of whether or not instrument constituencies are inherently good or bad things, they simply are. If anything the authors want to put the constituency issue on the radar because quite often it is overlooked.

Second, this is gentrification by another name

This squares nicely with my earlier analysis of the gentrification of #w2p (again, see: The Gentrification of #w2p); where obviously the policy instrument in question was social media.

Third, this offers us some predictive capability

Namely, that this is likely to happen (or is already happening) in other hot emerging policy areas/instruments right now:

  • Open Data
  • Policy Innovation
  • Design Thinking
  • Behaviorial Economics
  • Social Finance

Fourth, you may want to know how to spot members of the constituency

If you are looking to suss out whom might fall into a particular constituency it may be worth thinking about whether or not a given person more readily identifies with the policy instrument with which they work or the policy domain within which they work. In other words, is he an "open data" guy, or is he a "Treasury Board" Guy?

Fifth, its important to know who's who

Knowing who the members of the constituency are is important because the evidence indicates that these will be the people with greater influence down the road, they are likely people you want to cultivate relationships with now, before everyone else starts coming out of the woodwork.

The only trick is picking the right instruments. 

Not all innovation takes root. Not all new instruments mainstream. There was an early push on crowd-souring years ago. A lot of people took kicks at that particular can, its a tool that is still in use, but not nearly as widely as was originally thought.

On the (Seemingly) Partisan Nature of Public Sector Unions

Friday, September 11, 2015
by Nick Charney RSS / cpsrenewalFacebook / cpsrenewalLinkedIn / Nick Charneytwitter / nickcharneygovloop / nickcharneyGoogle+ / nickcharney

Can someone please reconcile something for me?

How can public service unions engage in (seemingly) partisan activities such as "jump[ing] into the third-party advertising fray for the current federal election campaign" while still purporting to represent the interests of the professional and non-partisan civil servants who make up their membership?

Consider the following the Professional Institute of the Public Service of Canada's recent press release entitled "Radio ads poke fun at Harper government to raise awareness of the impact of government science cuts and muzzling" (click to enlarge):

Or, if you prefer, listen to the ads themselves. Their permissibility seems to stand in stark contrast to the non-permissibility of similar actions by individual civil servants (See: On Non-Partisanship and Anonymity in the Internet Era).

Am I missing something?