Reading Manuel’s post, it became quickly apparent to me that it contains some real gems. Yet something in the overall editorial and logical flow didn’t quite click for me (reminded me a lot of my grappling with the Heifetz book). So it took me longer than I expected to get to a distilled version. It’s still far from perfect, and I’m sure I’ll come back to this topic in the future, but it’s a better baseline to work from.
Manuel’s piece covers several topics:
The distinction between accountability and responsibility
The accountability process
Reliable promises and predictable results
The distinction between consequence and punishment
Support & rescue
Scaling accountability for multiple teams
Pre-requisites for accountability
As you can probably tell, that’s A LOT to digest. I’m going to focus on 2&3 here which are the core elements, in my opinion. #6 is also key but already has good coverage here, here, here, and here.
What is accountability and how does it work?
Collaboration is a network of people making promises for delivering certain results to each other. The promises make the results possible, but can never guarantee them.
When collaborators repeatedly uphold their promises, trust is established.
When trust is present, people interact with each other without cautious defense mechanisms that drain part of their energy and the collaboration becomes more and more effective.
However, when collaborators repeatedly don’t achieve the intended results, trust erodes and can easily turn into resentment.
Accountability is a refinement of the collaborative social contract that’s intended to help all parties involved maintain trust: the collaborator making the promise authorizes the other party to evaluate the result they’ll deliver, and enforce a logical consequence for that result. The consequence was ideally agreed-upon when the promise was made.
The key things to note here are the clarity on the role that accountability plays in supporting effective collaboration efforts, and its directionality: accountability is initiated by the collaborator that’s making the promise, authorizing the other party to hold them accountable.
Another powerful lever in delivering predictable results that reinforce trust is making reliable promises in the first place, which reduces the need to utilize accountability. Part of it has to do with being able to accurately assess our own level of skill/capability in regards to the particular result that we’re going to make a promise about. A notion that traces back to Andy Grove’s Task-Relevant-Maturity and most likely even further than that. But the other piece of that puzzle is being mindful of the level of predictability of the context in which we’ll be operating (inversely correlated to its complexity) which can also put a strong constraint on the type of result that we should be promising:
I find the distinction between these 4 types of results: input, output, outcome, and intention; and the way they interact with the context extremely powerful. Recognizing the relationship between result type and context type seems to be missing from many of the conversations around goals setting, accountability and effective collaboration.
It’s funny and sad at the same time when a Dilbert cartoon comes to life in your company. As Brendan Schwartz, CTO of Wistia recently learned.
I’ve explored goal setting in the past and it’s definitely a topic that’s due for a more comprehensive post with my updated thinking on it. But I’ll keep this one short and more anecdotal. Schwartz’s piece is super short as it is so there’s not much summarization or synthesis to be done:
At least in Tech, it’s become accepted as conventional wisdom that people are most motivated when their goals are “stretch” goals — targets that lay just beyond the rational limits of what’s possible. But Schwartz found the opposite at Wistia. Setting stretch goals led to two counter-productive symptoms:
Short term thinking: stretch goals put the team in a continuous catch-up mode, desperately looking for ways to meet the (short-term) key results, often times, at the expense of values and long-term consequences.
Demotivation: consistent with some supporting academic literature, the Wistia team found that when everyone sets goals they know they can hit with hard work, it creates a cycle of positive reinforcement, keeping people motivated and marching forward. Motivated employees don’t just sit back on their laurels after they achieve their goals. They set new ones, exceed those, and expand their view of what is possible along the way. As time goes on, they accomplish bigger and better things.
Schwartz’s conclusion really hits the nail on the head:
Setting intentionally out-of-reach goals reflects a cynical way of thinking about human nature and motivation. It’s driven by a belief that people are lazy, and by default won’t be ambitious or creative or try to do more than they think is possible. Goals, therefore, become a way to correct for that laziness. This way, even if people fail, their output can still be decent.
There’s definitely some missed nuance here. What Schwartz is describing sounds more like “impossible” goals than “stretch” goals, which were always described to me as “possible but not probable”.
The challenge in this definition is that it’s a judgment call, to begin with, and assumes perfect ability to predict the difficulty of a goal which is a fool’s errand in a complex world. So we can certainly make the case for the “art of goal-setting” but we can also question the value of the practice to begin with.
Thoughtful behavioral design is a required prerequisite for the effectiveness of any organizational policy or program. Mindful use of defaults is a hallmark of good behavioral design. But not all defaults are created equal.
A team led by Jon Jachimowicz set out to conduct a meta-analysis on the effectiveness of defaults and summarized their academic paper in this more readable blog post:
Overall the team looked at 58 default studies with a total sample size of 73,675 participants. On average, they found defaults were a strong choice architecture tool, shifting decisions by 0.63 to 0.68 standard deviations: in decisions where there are two possible options, the option that is preselected is on average chosen 27 percent more often than the option that is not preselected. Given that other behavioral interventions tend to shift decisions by 0.2 to 0.3 standard deviations, defaults, on average, were two times more effective. But averages only tell part of the story, since the effectiveness of specific default interventions varied greatly. From significantly more than average effectiveness to not effective at all.
They reflect an implicit endorsement from the choice architect.
Staying with the defaulted choice is easier than switching away from it.
They endow decision makers with an option, meaning they’re less likely to want to give it up, now that it’s theirs.
In their analysis, they found that studies that were designed to trigger endorsement or endowment were more likely to be effective. The other aspect that impacts the effectiveness of default is the intensity and the distribution of the decision makers’ underlying preferences. When decision makers care less about a particular choice, a default may be more persuasive in swaying their decision. Likewise, when preferences within a population are more varied, such that some people may have preferences that align with the default, but many people may not, then a default may be less effective.
Yet, this is exactly what teams in modern organizations lack. Imagine trying to build a great theater ensemble or a great symphony orchestra without rehearsal. Imagine a championship sports team without practice. In fact, the process whereby such teams learn is through continual movement between practice and performance, practice, performance, practice again, performance again. — Peter Senge, The Fifth Discipline
A key enabler of greater organizational impact is the ability to discern between learning and performing. That distinction helps organizations evaluate their existing systems and programs and identify places where this distinction is not respected. As is often the case in their performance management system (reviews, compensation changes, etc.).
The first step that organizations that recognize this tension tend to take, is to decouple their “performance management” and “professional development” systems, both procedurally and temporally, so conversations about “how am I doing?” are separated from conversations around “how can I get (even) better?”.
While it is a step in the right direction, allowing them to at least capture the upside of “professional development”, the downside of “performance management” remains. Finding an alternative, or dare I say, completely letting go of the latter, requires revisiting the core employer/employee agreement: “you are getting paid to deliver outcomes”, and its underlying assumptions since it is the core tenet on which the notion that performance can be measured and managed is based.
Collaborate for social change has done some incredible work on this front, and summarized it in two reports:
As you may have gathered from the titles, the context is a little bit different: Their work aims to re-envision the relationship between funders/donors and non-profit organizations. But if we zoom out and abstract a bit, the core funder/non-profit agreement is identical to the core employer/employee agreement: you are getting paid to deliver outcomes. Therefore, many of their insights are transferable to our own context. The content below is a synthesis and organizing of content that I’m using almost verbatim from the two reports, with a few narrative sentences of my own where I will try to do my best connecting the dots.
The status quo paradigm in the non-profit space is called “New Public Management (NPM)” and is characterized by the ‘three Ms’: Markets, Managers and Measurement:
Markets — the creation of markets for social interventions helps to drive innovation and efficiency
Managers — social interventions must be overseen by people with training in professional management practice. Managers’ role is to identify what success looks like (strategic management) and to hold subordinates accountable, through performance management, for delivering it.
Measurement — Metrics must be created which identify what success and failure look like, and performance must be measured against these metrics
Nothing about this is non-profit specific. It describes the prevailing paradigm in the for-profit space as well. If you’re not convinced just replace “social interventions” with “corporations” in the bullets above.
However, this paradigm fails to factor in the fact that the world we live in is complex, across three key dimensions:
People are complex: everyone’s life is different, everyone’s strengths and needs are different.
The issues we care about are complex: issues — like homelessness — are tangled and interdependent.
The systems (organization) that respond to these issues are complex: the range of people and organizations involved in creating ‘outcomes’ in the world are beyond the management control of any person or organization.
The alternative, complexity-friendly paradigm requires working in a way that is human, prioritizes learning and takes a systems approach (HLS for short). See the full comparison table at the end of this post, but the key differences are the following assumptions:
Motivation — Those doing the work are intrinsically motivated to do a good job. They do not require ‘incentivizing’ to do the right thing. Instead, they need help and support to continuously improve their judgment and practice.
Learning and adaptation — Learning is the mechanism to achieve excellent performance and continuous improvement. Learning comes from many sources — from measurement and analysis, and also from reflection on the sensemaking and judgments we make every day in situations of uncertainty. This new paradigm views learning as a feedback loop which drives adaptation and improvement in a system.
System health: quality of relationships — outcomes are created by people’s interaction with whole systems, not by particular interventions or organizations. Funders and commissioners working in this way take some responsibility for the health of the system as a whole, because healthy systems produce better outcomes. They take a system coordination role. They invest in network infrastructure which enables actors in the system to communicate effectively; they invest in building positive, trusting relationships and developing the skills of people who work in the system.
People who work in a way that is informed by complexity use the language of ‘being human’ to describe what they do. This means:
Recognizing the variety of human strengths, needs, and experiences.
Building empathy between people — so that they recognize, and seek to act on, the emotional and physical needs of others.
Using strengths-based approaches — recognizing and building on the assets (rather than deficits) of people and places
Trusting employees to act on their intrinsic motivation to help others and get better at what they do.
Managers talk about ‘liberating’ workers from attempts to proceduralize what happens in good human relationships, and instead focus on the capabilities and contexts which help enable these relationships. They talk about providing support that is bespoke. For leaders and managers, being human means creating trust with and between the individuals, teams, and departments. Trust is what enables leaders to let go of the idea that they must be in control of the support that is provided using their resource.
People working in this way also speak about learning and adaptation. They describe how their work is not about delivering a standardized service, but rather that it is a continuous process of learning which allows them to adapt to the changing strengths and needs of each person with whom they work. Budgets and salaries and thought of as resources to enable organizations to learn and improve. They are not purchasing services with particular specifications, they are funding the capacity to learn and adapt to continuously improve outcomes in different contexts. This challenges traditional, narrow forms of accountability based on targets and tick boxes. To meet this challenge, organizations are recognizing the multiple dimensions of accountability, and exploring who needs to provide what kind of account to whom. This process involves dialogue, not just data.
Finally, people working in this way recognize that the outcomes they care about are produced by whole systems rather than individuals, organizations or programs. Consequently, to improve outcomes, they work to create ‘healthy’ systems in which people are able to coordinate and collaborate more effectively. From these organizations, we have learnt some of the characteristics of the ‘healthy’ systems that produce good outcomes, and the System Behaviours that actors exhibit:
People view themselves as part of an interconnected whole
People are viewed as resourceful and bringing strengths
People share a vision
Power is shared, and equality of voice actively promoted
In essence, the book is a collection of short, blog-size chapters each covering a different aspect of how they run Basecamp which together creates a fairly clear picture of the overall philosophy and the experience of working at basecamp.
In full disclosure, they had me at page 27 (of 234):
It begins with this idea: your company is a product… but when you think about your company as a product, you ask different questions: Do people who work here know how to use the company? It is simple? Complex? It is obvious how it works? What’s fast about it? What’s slow about it? Are there bugs? What’s broken that we can fix quickly and what’s going to take a long time?
I’ve always used this analogy and seeing the authors use it was an early indication that I’m likely reading a book by kindred spirits. My only nit about the book is its built-in marketing: the unusual and not-so-visually-appealing cover design (above) and the mouthful, not-so-catch title. Don’t let them discourage from reading this otherwise great book.
Paying homage to the agile manifesto, I decided to personally rebrand it as “The Calm Company Manifesto” and summarize its key points in the form of “x over y” statements:
Work & Life
Mutual give and take between work and life over life gives, and work takes
This one is going to be a little ranty. I think that’s ok 🙂
Not getting a seat, or a big enough seat, at the (executive team) table, is a common complaint I hear from many of my peer HR leaders and certainly a challenge I’ve grappled with myself. Granted, often times an interpersonal or a personal development gap for either CEO or HR leader (or both) is a contributing factor. But I also believe there are more systemic issues that are holding us back and must change. These are my Top-3:
Structure — it’s almost always the case that no matter how large the executive team is, no matter how large the company, all but a single member of the executive team is focused on the stuff the business creates (product, engineers, ops, sales, marketing, etc.) and only a single member is focus on the people who create the stuff. When stuff outnumbers people so overwhelmingly, people will rarely get the attention they deserve. Furthermore, much of the work that falls under the HR function is not really people-centric. It’s bureaucratic/administrative/compliance “organizational tax”, that yes, somebody has to do, but does it really make sense to roll it under the same person responsible for figuring out how to increase what this collaborative effort is capable of accomplishing? As Ram Charan compellingly argued almost 5 years ago, it may finally be time to split HR into two organizations led by two different executives: HR-LO (leadership and organization) and HR-A (administration).
Faux science — from MBTI-based hiring to anonymous 360 feedback surveys, so many organizational practices today seem to follow the 1944 Office of Strategic Services (CIA’s precursor) Simple Sabotage Field Manual rather than what decades of scientific research in psychology, sociology, and neuroscience taught us about the human condition. Liz Ryan does a very good job driving this point home in How Junk Science Set HR Back Fifty Years. There are so many wrong defaults in the “how to run an effective organization” (fictional) manual. Grounded skepticism towards existing practices, testing their underlying assumptions and ruthlessly eliminating/replacing practices that don’t really move us forward all have to be part of the solution.
The Dunnig-Krueger effect — Dunning–Kruger (DK) effect is a cognitive bias in which people of low ability mistakenly assess their ability as greater than it is. It is particularly strong when people have some experience with the ability (non-professional drivers are more prone to it than people who’ve never driven); and when the ability is more directly tied to their identity (most of us are probably worst kissers but better dancers than we think). When discussing topical matters that don’t pertain to their own function, executives are usually fairly immune to the DK effect. If I’m the VP of Sales and this is an Engineering issue — it should be relatively easy for me to recognize my own lack of expertise in Engineering and defer to the VP of Eng’s expert opinion. But all execs lead teams, and consider being a good manager an important part of their role (identity) so when it comes to discussing people issues, everyone thinks they are an expert.
As noted above, while I have some ideas for combatting #1 and #2, I haven’t come across anything that offers a good way to remedy #3 just yet.
In full disclosure, I have not read the book, just an article that was highlighting this particular piece of useful content from it.
Mistakes are a natural part of life. We all make them. And I don’t think any of us would want to live in a culture where no mistakes are tolerated. There has to be a path for recovering from mistakes, and the difficulty of that path should probably be proportional to the harm caused by the mistake.
Amy Rees Anderson offers this nifty framework for properly apologizing for a mistake. A key component of recovering from one:
Admit — I made a mistake
Apologize — I am sorry for making the mistake.
Acknowledge — I recognize where I went wrong that caused the mistake.
Attest — I plan to do the following to fix the mistake, on this specific timeline.
Assure — I will put the following protections in place to ensure that I do not make the same mistake again.
Abstain — Never repeat the same mistake.
Going to keep this one handy. Just like any other human, I’ll probably need it sooner rather than later.