Leaflet #113

1. Matrix management for clarity 2. Standing out vs. standing in with AI 3. Immodest proposal: don't do meta-work

1. Matrix management: using solid and dotted lines so everyone has a manager

In a traditional reporting line, if I report to you, you’re the person I go to to meet my needs. The manager is a one stop shop. In a growing organization, you’ll often have a project that doesn’t align tightly with the shape of your org chart. You need people to work on something that isn’t in the purview of their official manager. That manager can’t be a one stop shop for this person, while they’re working on this project.

In effective matrix management, you maintain clarity about whom to go to for what. The “whom” will be more than one person, not solely the “official” manager.

You’re going to have a dividing fork in assessing who meets needs of each sort. If I have needs related to this project, I go to Person A. If I have needs related to that project, I go to Person B. And if I can’t decide, I know exactly who I go to when I can’t figure it out (could be Person A, Person B, or a whole other Person C, depending on your context).

Organizations that adapt well in these situations define what a solid line looks like in a reporting relationship and they define equally well what a dotted line looks like in that relationship. When they can’t have a one stop shop manager, they ensure there is a backstop manager.

The particulars depend altogether on the context. No matter how those particulars are arranged, the individual is fully managed. There’s never a situation where they feel like they don’t have a manager they can go to for needs or guidance.

-ben

standing out vs standing in with AI

You can rapidly use an LLM to create a useful, credible “stand in” for an otherwise expensive thing you need. You can often do this in one shot: you write a prompt, within minutes you get a working draft of the thing. If, however, you need to “stand out,” you shouldn’t count on one shot. You should plan to engage in a thoroughgoing exchange, make a ton of edits, and, if you’re new to the domain where you’re making this thing, building skills and judgment of your own.

Succeeding in an exchange like this, where you operate at the edge of your own skills in tandem with an LLM, is a skill in itself.

Two illustrations of “standing in”:

TASK 1: Last summer, I ran a hiring process where we explicitly encouraged candidates to use LLMs in their late-stage performance tasks. Candidates who had made it through screeners and a couple 1:1 interviews had to respond to several complex prompts in less than a couple hours.

When we got back the responses, it was striking how … almost all of them sounded exactly the same. It was clear that most candidates pasted our prompts into GPT 4o (the default free model with biggest brand recognition at the time) and then pasted the output from GPT into a google doc and hit send.

This made our job as reviewers a lot easier. None of these copy-paste candidates stood out. They all blended together. To be clear: their pasted responses weren’t “slop.” The responses were thorough, smart, and clear. The writing had no obvious mistakes. But the responses were all basically the same.

TASK 2: I’ve recently gotten ~obsessed with native trees and their benefits for my beloved, precarious New Orleans landscape. I’m doing a pro bono project in conjunction with GW Carver High School – we want to plant an “Alumni Forest” where each graduating class since 1958 has its own tree and each forthcoming senior class will plant a tree of their own.

Long story short, I used Claude Opus 4.5 to generate a 3-D visualization of my draft design for the forest and the small park that it will sit within. It’s a little buggy, but you can “walk” through a virtual version of the alumni forest, toggle between stages of tree maturity, look at the design from birds-eye view, ground level, and any angle between. Having a landscape architect build this would likely cost hundreds, even thousands, of dollars. I got it for (close to) free in three minutes.

This visualization is a stand in. It’s a rough draft that accelerates progress toward a real thing in the world, that will be built with community and idiosyncrasy, that will ultimately stand out.

-eric

immodest proposal: don’t do meta-work

In this piece, former Rockefeller Foundation exec Zia Khan talks about all the “meta-work” he did at one of the world’s largest philanthropies. Meta-work is the planning to have a plan. It’s the strategic, theoretical, back-to-first-principles writing and deck-building. Sometimes it takes the form of “stakeholder engagement”.

My immodest proposal to you - don’t do this kind of work. If it must be done, assign it to an LLM. You, human, agentic, idiosyncratic you - you build a draft of the actual thing. Run a pilot. Bring a rough version of your plan into the ring, let it take one on the chin, then decide what’s next.

This is my advice for most sectors but I think I offer it with most enthusiasm if you work in some kind of social service sector. The thing you want to do or offer is probably going to be useful to the community your organization exists to help. Instead of meta-work, do some tiny amount of that! See if it’s as helpful as you think. At the worst, it’ll have no effect. You pay the opportunity cost of the modest time and cash you put into the experiment.

Worth noting: you woulda paid that opportunity cost anyway, sitting at your desk editing the footnote on the white paper about the experiment.

This advice is coming from a personal perspective. The kind of perspective you have when your eyes are squinty because you are also gnashing your teeth and gripping a tabletop with both hands.

Here’s what I mean: A handful of years working in philanthropy made me a little bit of a crazy person. Zia Khan’s experience there is common. There’s so much (high-salary) talk about social change and deep inhibition about the walk of social change. As far as I can tell, the teams that make the biggest demonstrable difference absorb the bruises and reputational nicks from their fearful peers and actually give the money away. They actually build stuff. They don’t coast on “convening power” or “building the ecosystem” or whatever other jargon-y camouflage is available for “spending the money on talking with other people who aren’t spending the money about how we might someday risk-lessly spend the money, meanwhile spending the money on ourselves.”

One reason I have deep respect for the folks at Coefficient Giving - they look at the proposed use of almost every dollar and ask: “Is this going to do more good than just giving it, straight up, as cash to someone in need?”

-eric

COMPELLING QUOTATIONS

Writer Sasha Chapin on misguided reaches for happiness:

The most interesting finding in human psychology, ever, is that basically all of us are born with the wrong intuition about how to be happy. We think that doing enough tanha, enough grabbing, will please us. But actually, grabbing is the source of something like 90% of our unhappiness. It’s a basically dissociative reaction to reality which creates a sense of temporary dissatisfaction, like putting on tight shoes so that later your feet will feel good again. Skip it altogether, as much as possible, if you want to be at peace.

Journalist Jerusalem Demsas on the price of expertise:

Delegation to an independent authority is a commitment device not to allow short term political pressures to influence macroeconomic policy. But the commitment device is straightforwardly anti-majoritarian. The risk is that policymakers will take steps that the public doesn’t like.

Observer Chuck Klosterman on old selves:

When you start thinking about what your life was like 10 years ago--and not in general terms, but in highly specific detail--it’s disturbing to realize how certain elements of your being are completely dead. They die long before you do. It’s astonishing to consider all the things from your past that used to happen all the time but (a) never happen anymore, and (b) never even cross your mind. It’s almost like those things didn’t happen. Or maybe it seems like they just happened to someone else. To someone you don’t really know. To someone you just hung out with for one night, and now you can’t even remember her name.