MapBrief™

How well do you know the town you call home?

After twenty-six years in Denver, most of it tucked up in its northwest corner, I suspected my own honest answer would be “not very”.

So when I decided to take a personal sabbatical of sorts this Spring to ponder a mid-career refactoring in the brave new world of AI, I reflected that I needed to get a bit agentic myself about “touching grass” in my home city and remediate the flawed familiarity with the geography around me that had been stunted by the routine spatial habits of everyday life.

The Project

    • jog a mile in each of the city’s 78 statistical neighborhoods
    • include a city park in the route (only 4 neighborhoods had no park)
    • access neighborhoods using a combination of public transportation and bicycle

Thoreau famously travelled widely in Concord; it was time for me to travel widely in Denver.

Denver’s 78 neighborhoods (for an interactive map with jogging routes click here)

Despite the cumulative distance equaling three marathons, the physical component wasn’t really the point. Trust me, no one in notably fit Colorado is impressed by 11-minute miles. No, this was about being intentional in seeking out the inconvenient and unfamiliar other sides of town that were never “worth” my time. We are told that searching out the new and novel in distant lands is worth the expense in order to acquire “broader” perspectives. What if a similar un-narrowing was available close to home for next-to-nothing?

The Cognitive Consequences of Our Spatial Habits

The research is clear: spatial navigation is deeply entwined with cognitive ability and memory. We know that these functions are co-located in the brain’s hippocampus: note the common dementia symptom of “wandering”. But how are we to understand how the brain is influenced by our everyday routing decisions?

The foundational paper is Edward Tolman’s “Cognitive Maps in Rats and Men” that laid out the case that our habitual travels create linear “strip maps” of our surroundings rather than a broader, more comprehensive understanding. Observing rats in a maze, repeating successful routes “works” until the maze changes and confusion sets in. The unchanging everyday routes inhibit our internal Waze-like ability to re-route on the fly in the face of the unexpected. Hence, our dependence on…Waze, Google Maps, et al deepens. Subsequent research expanded on this strip map idea to show that we store routes as corridors but are bad at understanding what is adjacent to the corridors and indeed moving between corridors.

Nobel Prize-winning research (2014) points to an even more mind-blowing relationship: the same grids of cells in our brain on which spatial memories are encoded are also used for organizing non-spatial memories, critically providing temporal coherence and the ability to “replay” memories. Clearly in the realm of our brain architecture, spatial couldn’t be more special.

You see a collage of selfies, I see spatial anchors of memory

* * * * * *

On July 3rd I was just another jogger in Washington Park doing the loop and taking advantage of the cooler early morning hour. Sure, finishing all 78 neighborhoods in 66 days had the psychological satisfaction of completing a goal, regardless of its scope of ambition. But maybe the win was a fresh skepticism of the algorithmically optimized routes that dictate my daily movements. And in being more intentional about the inefficient detours and the goings out-of-the-way for their own sake that would fire a few more grid cells in the brain that would otherwise be dormant. After all, the science points more clearly than ever that who we are is where we’ve been and a coherent memory of self is not the product of frequent flyer miles or passport stamps but rather a richer, broader geographical understanding of the everyday places we call “home”.

— Brian Timoney

What gets measured gets managed.

I made a pilgrimage to the ruins of the birthplace of “what gets measured gets managed.”

The Midvale Steelworks in North Philadelphia is where Frederick Taylor and his stopwatch laid the foundation for “scientific management” at the turn of the 20th century. Notable too was his assistant there, Henry Gantt, who developed the eponymous project schedule chart. The location stopped producing steel over a half century ago, but the dream of precise productivity measurement continues to haunt our 21st century workplaces.

We in tech never really put away Taylor’s stopwatch, whatever the acknowledgements of its shortcomings when it comes to “knowledge work”. Because management all too often falls back on countable things: hours (preferably billable, or tax deductible), lines of code, pull requests, etc. And we all play along with the kabuki theater of KPIs during performance review season, pretending we all aren’t falling victim to Goodhart’s Law in the meantime.

And now AI has come along and proven very adept at producing countable things–lines of code and pull requests–around the clock(!). And it even spawned its own new countable thing–tokenmaxxing–initially embraced eagerly by management that alas, in record time, fallen victim to its own incentives.

Danger for Those Specializing in the Easily Measurable

Among the recent wave of tech layoffs, consider Cloudflare’s CEO on laying off 20% of its workforce despite profitability:

AI isn’t coming for builders or sellers, but it is coming for measurers. Tireless, independent, efficient and available, AI systems can now measure an organization with a level of objective detail and precision that was previously impossible even for the best employees.
…
The vast majority of those we laid off last week were measurers. We cut middle managers across the organization because AI allows us to have more direct reports per manager while still measuring and mentoring our teams effectively.

ClickUp CEO Zeb Evans had even more detailed and interesting thoughts about his own company’s 22% cut in headcount:

We must reward people that create productivity accordingly. This aligns incentives on both sides. Plus, in a world where your best people create 100x impact, you can’t afford to lose them.
…
Compensation bands of today should be thrown out the door. We’re introducing $1 million cash/year salary bands with a path available to nearly everyone in the company if they produce 100x impact by creating or managing AI systems.

Million-dollar salaries–sign me up! But a quick question here: I’m seeing this phrasing “100x impact”–what, precisely, is “x“?

Triumph of the Qualitative?

If AI has taken over the old metrics–hours worked, lines of code written, et al–how are we humans to be evaluated in the workplace?

The early consensus seems to be centering on “taste” and “judgment“!

Given my experience in technical/engineering teams for over two decades, I can confidently state that our management structures are wholly unprepared to evaluate tech workers on the basis of such qualitative characteristics such as “taste” and “judgment”. I mean, have you ever heard these would-be arbiters of taste discuss their favorite films? Is this really the “x” we’ll dutifully build a new set of KPIs around?

No, my guess is we’re still a far way off from replacing Taylor’s stopwatch because the ‘objectively’ quantitative is the comfort food of our world of zeroes and ones. We will permit its hollow ruins to stand if only as a monument to our own uncertainty of what to replace it with.

— Brian Timoney

Last year I ran into the always-incisive Will Cadell and we immediately started discussing a favorite hallway-track topic:  how to effectively sell Geo.

ME:  “If I had to do it all over again, I wouldn’t even bother with the web but just sell PDF maps to the Oil & Gas industry with big red arrows that said ‘Drill Here’.  At least with a PDF they know exactly what they are buying.”

WILL:   “They are not buying the PDF; they are buying the arrow.”

Cue An Epiphany.

We in tech love latching on to metrics: map a billion points in your browser, access petabytes of imagery using a shiny new platform, etc. etc.

But so little conversation about the arrows our customers actually want. 

Instead, we jam a billion points into their browser and wish them luck in finding whatever answer we thought they were looking for.  Sure, it is better than the old days when we actually handed customers hard drives of data with an invoice (!), but not by much.

To torture a different analogy, the customer is looking for a needle and we respond by trying to sell them a bigger haystack.

I get it: each client’s arrows and needles are too unique to build a business that can truly “scale”.   Or maybe AI will soon bequeath us an All-Purpose Answer Machine.  In the meantime, perhaps our time is best spent not heads-down grinding on the latest GPU-accelerated ways to push more data at our end-users but maybe circling back in a quiet moment and asking “wait, what was your original question?”

— Brian Timoney