Bits and Bobs 11/27/23

When you have a massive tangled bramble of magic and special cases, it's impossible to think about it in broad, coherent ways.

Every single feature is a special case; every incremental tweak or addition tends to add yet another special case.

When you have a conceptual model that captures all of the real world semantics of the system today (and also the expected semantics for the next few years) that's a game changer.

Now, incremental work can align itself with that model, cleaning up what pre-existing parts it touches, and laying down new aligned components.

Each incremental bit of work you do reduces the amount of overall magic, just a bit.

Over time, the system will naturally get more coherent, in an emergent way.

At some point a customer will ask "Hey can you provide us with a knob to do X", and they'll describe something very similar to how the internal system actually works.

At that point, it's a no-brainer to expose the knob (perhaps in the basement, not for non-savvy users), because you know it's a fundamental long-term semantic and not some random implementation detail.

If you have a feature in your product you think is ugly, but your users tell you it's beautiful, then it's beautiful.

With user diversity and/or scale it's not possible.

That's why the ideal is a "pre-assembled lego set".

Spaghetti code accumulates where the value flows through in a system.

For the user, mucking with it is all downside and little upside.

Spaghetti code in a context tends to only ever accumulate.

You can't hope that a customer moves their spaghetti code over to your platform, or to create a clean platform where users won't need to ever do any spaghetti code.

The play is not to replace spaghetti code, it's to be the place where a customer's spaghetti code lives.

It's OK to offer customers suggestions in case they think you're smarter and want your opinion.

But never assume that you're smarter and force them to take your suggestion.

In complicated circumstances (balancing lots of different sub-components of different types) it can be considerably more difficult.

In those situations, a thing that looks like a no-brainer might actually be a bad idea.

Situations that "should" be simple can sometimes be accidentally complicated; a kind of "gerrymandering".

Making it so the important decisions are easy to make well with few moving parts is a form of paying down debt.

Game-changing ideas are basically impossible in this situation.

A game-changing idea fundamentally will have at least some concrete downside from the status quo, and might have upside (and sometimes that upside turns out to be transformative).

A few places this dynamic shows up:

Large companies with bottom-up cultures

Peer review of academic papers

A better goal is: "no one thinks this will kill us and at least one person thinks it will be actively great"

In an environment that is amenable to experiments, this is a much better goal and more likely to find great things that work.

You can also invest time and effort to build guardrails to reduce the cost/danger of the average experiment, allowing you to do more of them.

In those typical situations, you want to consider a lot of options, carefully choose one, then focus on it.

But in environments where there is tons of value everywhere (e.g. after a general-purpose, game-changing technology becomes widely available), that advice is wrong.

In those cases, the danger isn't that you pick the wrong thing, it's that you don't pick anything.

In those cases, it's easy to get paralyzed figuring out which option will have the highest value.

It's like Buridan's ass, which had two equally good bales of hay that it couldn't decide between, so it died of starvation.

Part of the problem in these environments is that you don't know which option will have the highest expected value.

But no matter which option you pick, you'll almost certainly pick one that creates a lot of value.

Knowhow (intuitive, situated knowledge borne from experience) is extremely rich... but impossible to transmit to others at even a small fraction of fidelity.

If you have to convince many others before you can do anything, you'll do nothing.

It's liberating to be able to take small, experimental actions yourself.

Then, ideas that turned out to work become a no-brainer for others to double-down on once they're shown to be successful.

A real world example, a second hand story from Google Maps.

When viewing a place page, there's a button to get directions, with a big obvious icon.

The text beneath the icon used to be how long it would take to drive there, if you left right then: "15 minutes"

When experimenting they found, somewhat surprisingly, users seemed to like it better if the text instead said "Get Directions".

But this improvement was an illusion.

The "user satisfaction" proxy was CTR, and CTR did indeed go up when the text was changed to "Get Directions".

But that was not because users understood the button better.

Instead, it was because a class of users who previously had their needs met without clicking the button ("Do I need to leave now?") now had to click through to get their information.

A quick 2x2 of two dimensions:

Personal vs Shared

Mundane vs Transcendent

Turbulent (Personal / Mundane)

The default state.

Random, reactive.

Never getting time to execute on any thread of creative effort for more than a moment, never knowing what's next.

Nothing of value emerges from this state.

Clockwork (Shared / Mundane)

The common state for adults in Serious Work.

Everyone synchronized to an external clock.

Everyone subverting their own rhythm to the centralized clock of the machine of society.

By being in sync, things that are impossible to do alone become possible.

Always knowing how many minutes you have until your next time-based commitment / meeting.

One eye on the clock, constantly.

Great for execution on plans.

Gives you a background metronome and structure to slot work into, a default forward momentum.

"I have 45 minutes to get A, B, and C done. I need to get started or I'll never get it done before the next meeting".

All of the synchronization and pinned down plans make it hard to adapt.

Many good things emerge from this state.

Flow (Personal / Transcendent)

Entirely in sync with your own internal creative energy and potential.

Disconnected from time and those around you, fully in flow state.

Only possible to get into this state when the muse strikes.

Being ripped out of this state feels like losing a limb.

All great things emerge from this state (or the next one).

Scenius (Shared / Transcendent)

Multiple members of a group in flow state together.

Lost from time and the surrounding system, but together in creative potential.

Massive ripple effects through time and space emerge from these acts of creation.

The highest possible state.

The most important things ever created come from this state.

Transitions between states are very, very hard.

The momentum of clockwork sometimes can give you a "kick" of energy so when the muse hits you you can go.

One way to make sure you have momentum when the muse hits is to have a bunch of pent-up ideas you're just waiting for some time and space to do, so when you finally get the time, you can just go.

The different states are more or less frequent. A ballpark guess of how many hours of humanity's time is spent in each:

Turbulence: 60%

Clockwork: 30%

Creative: 9.9%

Scenius: 0.1%

We can get better or worse at putting ourselves in the right state at the right time.

How can you get out of Turbulent state more often? How can you get more time for Creative?

How can we nurture the potential for Scenius?

You have to get everyone to point in the same direction at the same time.

The cost gets super-linearly more expensive with the size of the group to align.

When there is an existence proof in the world to point to, coordination gets orders of magnitude easier.

"We want to do something like that, but with these tweaks."

Everyone can see with their own eyes that something similar is viable.

Once you have some success with a group, it's easier to keep it going than to do something new.

Static friction is higher than rolling friction.

Once you get going, you can factor out the alignment into external structure to make it easier to stay aligned.

Process and infrastructure are examples of this structure.

A kind of stigmergy for coordination.

You can think about processes like cached coordination results.

Phase transitions are transformative and game-changing.

As a person, when you're at peace, you're in a form of dynamic equilibrium.

Being at peace is being in balance, ready to move decisively into a new phase when the opportunity presents itself.

The former is approaching stasis: death.

The latter is poised to capitalize on game-changing opportunities when they present themselves.

Both can look almost lazy to outsiders, but the latter can create transformative value.

What would you do differently in the short- to medium-term depending on those different extremes?

If the answer is "nothing", then don't do the analysis, just do the thing!

This is especially true if the downside risk of doing the thing is small and capped.

If it's a low cost no-brainer, the bar to clear should be very low.

The opportunity cost of analysis, of bringing every last person along, is in practice the largest component of execution cost in large organizations.

The magic doesn't come from any particular sub-component.

It emerges out of the connections between the parts.

Where's the magic? The whole thing!

This is one of the reasons that it's hard to analyze an ecosystem.

"This extremely popular SaaS platform with a big ecosystem is a crappy product that no one seems to like, and yet it has accelerating momentum. Where's the magic?"

Don't tear down what you hear from others; steelman it.

Not "what this person is missing is...", but "what this person is seeing is..."

Fundamentals are hard and nuanced and slow.

In the end, optics don't matter. Fundamentals are everything.

You don't know which ones will turn out to be mighty oaks, and which ones won't survive.

But they all start out as seedlings, and seedlings are cheap to nurture.

Seedlings just need some shade and patience.

As time goes on it will become increasingly clear which seedlings have the most potential, and doubling down on them will be a no-brainer.

It's "humbly seek disconfirming evidence and absorb it into your mental model, and don't waste time chasing down the illusion of certainty".

People don't want to look too closely at their own imperfections; it's threatening and scary and they'd rather not do it.

A narrative trick: make a mirror that looks like a painting, so people can engage with it more deeply.

"Hey wait a second, this painting actually kind of reminds me of us!" / "Really? What a crazy random happenstance!"

Best practices have to be grown in place, over time.

You can't duct tape cultural practices willy nilly; they won't fit, and are often incompatible in non-obvious ways.

The only thing to do is to set a direction and over time arc towards it, making tradeoffs between incompatible components when necessary.

It's not possible to have all three of the broad bottom-up entrepreneurially of Amazon, the quality of Apple, and scale.

A successful experiment is an existence proof of viability to double down on.

Don't judge teams by how many small, safe experiments fail.

Judge teams by how many experiments succeed and can be built upon.

Don't try to pre-judge which small, safe experiments will work out. Just do them.

You get to be an idealist about the parts that won't kill you.

Even when it's cooked well, it's merely good, rarely great.

The vast, vast majority of turkey dinners are dry and bland.

There are a lot of situations in life like roasting a turkey: very hard to do well, and even if you do there isn't much upside.

If you aren't roasting the turkey for some other reason (connecting with family and society over a shared tradition) then maybe don't bother?

That is, scale, efficiency.

When the system is mostly unknown and fast changing, you want community gardening.

That is, resilience at the level of the system and local experimentation.

If you look for the magic in any individual component, you won't find it.

And yet the emergent result is extraordinary.

He relayed a story about an octogenarian friend of his.

His friend doesn't own a smartphone or use technology.

When Max picked him up from the airport, Max navigated home using Google Maps on his phone.

His friend looked at the phone and asked, "Is this AI?"

Perhaps the answer isn't as straightforward as we think.

That's kind of a bummer: we aren't special!

A few days ago my 4-year old told me she was sad because in a given situation she realized she wasn't special.

I told her: "It's OK, honey, in the grand scheme of things, none of us are. But that doesn't stop us from doing special things!"