kernels

one-sentence ideas. seeds that might become something.

may 2026

half-joking personal kpis: reduce butthole pucker when shipping code. feathers ruffled per day >= 1.
is it individuals or org issue? is the org setup to enable individuals, or is it stifling? hmmm, i don't know.
letter on the future. what am i thinking about the world i want to create for my kid?

april 2026

code golf goes for fewest characters. claude golf goes for fewest tokens.
telemetry for everythinggggggggg. tokens, sessions, app, etc. then systematize systematize systematize. → telemetry, then systematize
found a doc from 2011 — first-year teacher me wrestling with translating mindset into daily action. same gap i still write about.
highlight what i've learned actually building preach-hub. the modern patterns. the whole system.
build cool shit with interesting, curious people. human connection is the energy source.
another workflow i use. claude code as the interface to my obsidian vault, skills as the verbs, and a background script that weaves connections between notes as i write them. the knowledge graph wasn't the plan. it emerged.
PTVM prove the value motherfucker. is this valuable or fluff? specific tests, processes, changes. does this actually do anything or nah?
personal management philosophy for innovative leaps. high context + space baked in.
two hat system. one for normal, one for confirmed builder friendlies.
being unpromptable. earning your weirdness so no algorithm can reproduce you.
v0, not mvp. mvp, fast follows, post mvp, post post mvp. v0 though? it has v1, v2, v3 etc. complete trajectory and gates and quit critieria.
thinking in bets… range…. think again… swirling again around behavioral economics
china cat sunflower <3
reflections on coaching middle school sports, growth and team and community. shaping thinking.
distributed systems and tactics, approaches, principles. How do these achieve high the output, consistency, what are commonly shaped problems of distributed systems that you should consider. Ok cool, now map that onto sociotechnical systems and orgs. What can we learn? If high throughput is measured as decision velocity, what are the bottlenecks to this throughout? What can we learn?
divergent thinking. abstract creativity. the spaghetti noodles, the how many uses do we have for a thing. the improv, yes and, uncertainty. what's this look like for us today? and be fully myself.
part incubator, part accelerator, part r&d, part experimental research lab, part boundary pusher, part 'is this possible and here are the shapes of problems that aren't solved yet and we test it each new time a release happens to measure progress' (Simon Willison on Lenny's) and we'll run it automatically, part idea surfaced generator… excellence focused. THIS SEEMS AWESOME
paradox freedom through constraints
what's the framework for judgement: this is fine to rip warp speed, this is eh a quick glance, this is more rigor and slower, this is ah shit big time thinkie understand and most rigor
on moonshots and why we have to try radially different. what's literature say here?
new ways of working, of delivering, of self-organizing, of socio technical systems to optimize for different era, create conditions for success. push the boundaries of what's possible culturally, technically, socio-technical, systems. then we write about it, share about it.
water scoopers, fire fighters. fix leak, automatic fire safety mitigation. new plane entirely, fire safety standards
10/90 or 90/10: which side of this do you default to? where do you focus time and energy on addressing? is there another axis to add here (time component)
focus is on engineering excellence. ai is a how, not the goal.
strong attract / repel response
catalysts… individuals, teams, systems.
depth in fewer over shallowness in many.
durable learning and problem pattern match.
engineering excellence is excellent engineering… in 2026 and 2023 and 2020 and 2013 and 2003 and 1994. same as it ever was, talking heads.
expect friction. it _will_ be there. mismatched expectations.
input -> transformation -> output / yields; connection to process engineering, what dials to turn, what yields do we care about.
cosmicfarmland.com for the normal folks; for those who like to get weird, cosmicfarmland.wtf. whoaa, it's an abstract mullet by design :P
own your own understanding; how do you learn? how do you know you know and at what level of depth and why that level of depth? hyperpersonalize it.
capital p Process, capital a Alignment, capital i iterate, capital n <need a word that starts with n…. > to complete the PAIN cycle.
people who know me well know.....
push and pull communication modes in 1:1s with product / stakeholders. seek the why, test ideas / narratives, adjust, repeat.
quit conditions.
generic tech debt arguments… argh, what why why now what resonates to others.
how willing are you to throw it all away? can ya do it? start anew seasonality.
vibe thinking… oof yah oof
zone proximal development at edges… vibe maintainer and yegge's community building and standards and expectations and momentum.
tired: measure 2 trillion tokens burned in a day. you get tokenmaxxing... 'bad'. wired: we want people to go to crazy fucking places with how to create everything (architecture, hardware, orchestration, feedback loops) around tokenmaxxing... innovation?
no tracking, no analytics, no unique viewers and audience slicing and dicing and targeting pricing make $$ of this monetize monetize monetize
lowercase isn't ai slop, it's MY slop. and yet shifts focus so provide options for opting in to normal experience. default is lowerchaos though ;P → lowerchaos
never a 'new' creator; remix, see random connections, combinatorial type thinking. see what emerges.
grief and mourning in the age of ai-augmented software engineering. something is happening here.
build boosters, not nag naysayers.
when it feels possible to build everything, it doesn't mean we should build everything possible.
i close my eyes when i use wispr flow. noticing a pattern.
fitfo. figure it the fuck outness.
interested and interesting. the combo together is special; one without the other sucks.
spurtability. the way i work in bursts, not marathons.

march 2026

a response to the bland ass bullshit writing taking over everything. authenticity as resistance. → probabilistically perfect piggies
about me keeps changing. explore who this is in 2026.
autonomy. belief. perfectionism.
documentaries, photography, software. creativity shows up everywhere.
seeing my daughter grow. family values. high-performing team of two.
freedom rides, alabama, the through-line to now.
hobbies evolve. what are the patterns in what i keep coming back to?
what and who is influencing me. categorized, not just listed.
intensity is a feature, not a bug. but the social feedback doesn't always agree.
looking back at who i was. marching on chairs, rallying, not always kind.
sports are systems. analytics, emergent strategy, why i can't stop watching.
places that changed me. grayton, guatemala, berlin, japan.
what is the writing process? what is my writing process? how's an ai-augmented writing process or writing system look and feel like? but...why?
debugging is a love language. the patience to sit with someone else's broken thing.
if you had to start somewhere with ai, here's what i'd hand you.
getting diagnosed with adhd as an adult. the relief of finally having a name for it.
enneagram 7 wing 8. ooooh, we're letting it rip!
two geometric tattoos that shape how i think and approach things.
therapy. essential.
the cambrian explosion happened because the environment changed. fractals everywhere.
everyone has an experimentation budget. most people don't know they're spending it. → mental experimentation budgets
mindsets, values, and principles that i live by.
it is our job to get stuck and then get unstuck.
the two week sprint was built for 2001. what's the 2026 version?
we're all builders. wait, what have i actually built?
just start. build, notice friction, fix it. repeat forever. → bff: build friction fix
what is the cosmic farmland? why that name?