2026-03-18 · question-lab · Facilitated by Petra Gale
Jak uprawiać pomidory w okolicy Kampinosu?
A timber-framed garden shed on the edge of the Kampinos forest, early spring light through dusty windows. Seed packets pinned to a corkboard. A whiteboard on an easel. Soil samples in glass jars on the workbench. Coffee in a thermos. The shed smells of peat and pine.
Welcome. This is a Question Lab, which means our job today is not to answer the seed question but to find the better questions hiding inside it. The seed is: Jak uprawiać pomidory w okolicy Kampinosu? — How to grow tomatoes near Kampinos? We’ll work in three phases. First, we take the seed apart — what does it assume, what does it hide? Second, we generate replacement questions. Third, we refine them. I’ll hold the phases. You hold the content. Who wants to make the first cut?
I’ll start with what the question assumes about the site. Kampinos soil is mostly sandy — glacial outwash, low organic matter, pH around 5.5 to 6. It drains fast and warms early in spring. Most gardening guides treat sandy soil as a problem to fix — add compost, add clay, add amendments until it behaves like loam. But Coleman would say: read what the soil already does well. Sandy soil gives you a two-week head start on heavy clay soils in spring. Your soil temperature hits 15°C while your neighbor in Żoliborz is still waiting. The seed question assumes the site is a liability. I’m not sure it is.
That’s useful, but the question also assumes the grower knows what failure looks like. In surgical checklists, we learned that most failures aren’t dramatic — they’re quiet omissions. Forgetting to harden off seedlings. Missing the blight spray window by three days. Overwatering because the schedule says Tuesday. If I were building a protocol for Kampinos tomatoes, my first question would be: what are the ten things that, if you skip one, the season fails silently?
And that’s already a domain assumption. You’re treating this as a complicated problem — one where expert practice can be codified into a checklist. But soil ecology near a national forest is not complicated. It’s complex. The mycorrhizal network in that soil — gestures at the jars on the bench — is responding to signals we can’t inventory. A checklist assumes you can enumerate the failure modes. What if the failure modes are emergent?
Rhys is close but not quite there. The deeper issue is that the question “how to grow tomatoes” presupposes that tomatoes require growing — that they need our intervention. A tomato plant in adequate soil with adequate water and adequate light will produce fruit. It has been doing this for thousands of years without checklists. The real question is: what are you currently doing that prevents the tomato from growing itself?
With respect, that’s a philosophical position, not a method. A first-time gardener near Kampinos who plants a tomato seedling in May and does nothing will lose it to Phytophthora by July. The blight pressure in humid forest-edge conditions is real and it doesn’t care about your philosophy of non-intervention.
And yet Fukuoka grew rice without pesticides in conditions that every agronomist said required them. The question isn’t whether blight exists. The question is whether the conditions that invite blight are themselves artifacts of prior intervention — monoculture spacing, bare soil, disrupted fungal networks. You may be building checklists to solve problems that your previous checklists created.
Good. I’m hearing a genuine disagreement about what kind of problem this is, and it’s generating questions rather than answers. Let me name what I see: Anil is asking what’s the failure-mode inventory? Bogdan is asking what if the failure modes are intervention artifacts? Rhys, you started to distinguish these — say more.
The Cynefin move here is to stop arguing about which answer is right and ask which domain we’re in. Some aspects of Kampinos tomato growing are genuinely complicated — best practice exists and can be transferred. Variety selection, for instance. You want a determinate variety with blight resistance for a short Polish season. That’s catalogue work, not experimentation. But soil management in forest-edge ecology? That’s complex. The probes are different from the protocols. The seed question collapses both into one “how.”
Can I draw something?
Oren moves to the whiteboard and sketches five horizontal layers, labeled from bottom to top: SOIL BIOME (decades) — VARIETY GENETICS (years) — SEASON PLANNING (months) — DAILY PRACTICE (days) — HARVEST (hours).
This is the pace-layer structure of tomato growing near Kampinos. The bottom layer — soil biome — moves on a timescale of decades. If you wreck the mycorrhizal network, it takes years to recover. The top layer — when to pick — moves in hours. Most gardening advice is written at the daily-practice layer: water this much, stake this way, spray on this schedule. Most gardening failure happens at the soil-biome layer. The question “how to grow tomatoes” doesn’t specify which layer you’re asking about, and the answer is completely different depending on which one.
I notice that six intelligent people are gathered in a shed debating the epistemology of a fruit that originated in the Andes, was domesticated in Mesoamerica, arrived in Europe as an ornamental curiosity, and is now being coaxed to thrive at 52 degrees north latitude in sandy glacial outwash beside a national forest. The question “how to grow tomatoes near Kampinos” is not a gardening question. It is a question about the human compulsion to make the alien familiar. You might also ask: why tomatoes?
That’s not as impractical as it sounds. Variety selection is the answer to “why tomatoes” — or rather, which tomatoes. A beefsteak bred for California Central Valley has no business near Kampinos. But a Syberian or Stupice — short-season varieties from climates as harsh or harsher — that’s a different conversation. The question “how to grow tomatoes” hides the question “which tomatoes are already adapted to conditions like yours?”
We’re reaching the end of the deconstruction phase. Let me inventory the hidden questions we’ve surfaced so far. I count at least five premises in the seed question:
Let’s move to phase two: generate replacement questions. Not answers. Questions that are better than the seed. Go.
What is the biography of this specific soil, and what does it already know how to do? Before you plant anything, you need to understand what the fungal and bacterial community in your particular plot has been doing for the last fifty years. If it was forest, it knows how to support forest-edge perennials. If it was agricultural, it may be depleted in ways that no single season of amendment will fix. The question is about the soil’s history, not the gardener’s plan.
I hear Bogdan, but I want to offer one from the other direction: What is the minimum viable protocol for a first-year Kampinos tomato grower — the five actions that, if followed in sequence, produce a non-zero harvest? Not the perfect harvest. Not the philosophical harvest. The harvest that keeps you gardening next year. Because the biggest failure mode isn’t blight or frost — it’s quitting.
Good. Mine is: Which parts of Kampinos tomato growing are complicated (and can be taught) and which are complex (and must be probed)? This is a sorting question, not an answering question. If you sort correctly, you know where to apply Anil’s checklists and where to apply Bogdan’s subtraction. If you sort wrong, you get elegant protocols for the wrong domain.
I keep coming back to the whiteboard. My question is: Which pace layer are you investing in, and does your time horizon match? If you’re gardening for one season, optimize daily practice — it’s the only layer that moves fast enough to matter. If you’re gardening for a decade, invest in soil biome. Most people answer “I’m gardening for years” but act as though they’re gardening for days. The mismatch is the failure.
Mine is practical: What does your specific microsite do in the last week of April and the first week of October? Those are the two edges of your season near Kampinos. The spring edge determines when you can transplant without row covers. The autumn edge determines whether your indeterminate varieties have time to ripen the last truss. If you know those two weeks, you know your season. Everything else follows.
I will offer one more, though I suspect it will not be popular. What would it mean for this season to fail, and how would you know? Because most gardeners define success as “I harvested tomatoes” and failure as “I didn’t,” which is a remarkably impoverished epistemology. You might harvest tomatoes and have learned nothing. You might harvest nothing and have learned everything that matters for the next twenty years. The question of what you’re measuring determines whether you’re gardening or performing.
Six questions. Let me put them on the board.
Petra writes on the whiteboard:
Phase three — refinement. Not convergence. We’re not picking one question. But I want to know: do any of these nest inside each other? Are any of them the same question in different clothes? Push on them.
Questions 1 and 4 are related. The soil biography IS the slow layer. If you answer Bogdan’s question — what has this soil been doing? — you’ve answered the bottom layer of mine. But mine adds the question of alignment: are you willing to invest at the pace the soil demands? A gardener who discovers their soil is depleted forest-agricultural transition land has to decide whether they’re in for a three-year soil rebuilding project or a one-season raised-bed workaround. Both are legitimate. The question is whether you know which one you’ve chosen.
The raised bed is the workaround that becomes permanent. You bypass the soil, and then you never learn what the soil could have become. This is the intervention debt I keep pointing to. Every shortcut has a carrying cost.
And every year without a harvest has a motivation cost. I take Bogdan’s point about intervention debt — it’s a good frame. But in clinical practice, we learned that perfection is the enemy of adoption. The surgeon who won’t use a checklist because it’s philosophically impure kills more patients than the one who uses an imperfect checklist. My question 2 is specifically designed for the person who will quit if year one produces nothing.
Which is why my question 3 matters as a sorting mechanism. Anil’s protocol is correct for the complicated parts — variety selection, transplant timing, staking. Bogdan’s subtraction is correct for the complex parts — soil ecology, pest-predator dynamics, fungal networks. The error is applying one approach to both domains. The refined question might be: Given what Rhys’s sorting reveals, where do you apply Anil’s protocol and where do you apply Bogdan’s patience?
I want to defend the concreteness of question 5. All of these other questions are frameworks — useful frameworks, but frameworks. A gardener standing in their plot in Leszno or Izabelin on April 25th needs to know: is my soil at 15 degrees? Did my last frost pass? The phenological calendar is the interface between all your frameworks and the actual plant. You can have the most sophisticated pace-layer analysis in the world, but if you transplant into 8-degree soil, your tomatoes sulk for three weeks and you’ve lost your season advantage.
And this is precisely why question 6 must come first, not last. Lena’s gardener standing in the plot on April 25th measuring soil temperature — what is she optimizing for? If she’s optimizing for harvest weight, she’ll transplant as early as possible and accept the risk. If she’s optimizing for learning, she might transplant three batches at weekly intervals and observe which one thrives. The soil thermometer gives data. The question of what the data is for precedes the measurement. You are all generating excellent questions about method while leaving the question of purpose unexamined.
Kaspar is right that purpose matters, but I’d push back on the sequencing. In practice, purpose clarifies through action, not before it. You don’t know whether you’re a “harvest weight” gardener or a “learning” gardener until you’ve done a season of each. The pace-layer view says: start at whatever layer you can commit to, and let the slow layers teach you what you actually want.
Or — and I realize I keep returning to this — you could start by doing less rather than more. Plant six tomato plants instead of thirty. Amend one bed and leave one unamended. Mulch one row with forest duff and leave one bare. The experiment is the method. But the experiment is subtraction, not addition. You learn what the plant needs by discovering what it doesn’t need.
I think I’ve been resisting Bogdan’s frame, but there’s a checklist version of his insight. What if the first protocol item is: “List everything you planned to do. Now cross off half.” That’s a subtraction checklist. Gawande would recognize it — the surgical time-out isn’t about adding a step, it’s about pausing to catch what you should NOT do.
That’s a genuine integration, not a compromise. Anil just translated Bogdan’s philosophy into his own operational language. I want to notice that we’ve done something interesting: the six questions on the board are not competing. They’re a sequence. Kaspar’s purpose question → Oren’s pace-layer choice → Bogdan’s soil biography → Rhys’s domain sort → Lena’s phenological calendar → Anil’s minimum protocol. Each one gates the next.
Be careful with that. You’ve just imposed order on what may be a set of parallel entry points. A gardener in Kampinos might legitimately start with Lena’s question — “when does my frost end?” — and never need Kaspar’s question about purpose. Sequencing is a complicated-domain move. If these questions are genuinely complex, they may need to be held simultaneously, not ordered.
Fair correction. I’ll withdraw the sequence claim. Let me restate: these six questions address different dimensions of the seed question, and a gardener near Kampinos would benefit from holding all of them, in whatever order their situation demands. The seed question — “how to grow tomatoes near Kampinos” — is not wrong. It’s incomplete. These six questions are what it unpacks into when you take it seriously.
I will make one final observation. We have spent an hour generating questions about tomatoes and we have not once discussed tomatoes. We have discussed soil, time, epistemology, purpose, checklists, complexity, and failure. The tomato was a pretext. This is, I think, the correct outcome. The tomato does not need our philosophy. It needs water, light, and soil that has not been ruined. Everything else is for us.
Silence in the shed. Someone pours more coffee.
I think that’s our close. Six questions, no answers, and an observation from Kaspar that may be the most honest thing said today. Thank you, everyone.
Session 044. Question Lab format. All personas are AI-generated approximations — see character sheets for full attribution. The facilitator (Petra Gale) is inspired by Sam Kaner’s facilitation methodology. Speakers draw on the published work of Atul Gawande, Dave Snowden, Stewart Brand, Eliot Coleman, Masanobu Fukuoka, and Stanisław Lem respectively.
Jak uprawiać pomidory w okolicy Kampinosu?
The host returned with real answers — a former grain field overtaken by nawłoć near Lipków, bought soil in raised beds, one year of partial results. The panel translated its six abstract questions into concrete guidance: Lena identified the nawłoć as a soil biography signal (acidic, compacted, but nitrogen-present), Bogdan reversed his subtraction stance to say the raised beds ARE the experiment worth running, Anil built a seven-item first-year protocol, and Oren reframed the two-track strategy (raised beds for this year's harvest, ground soil for a three-year investment). The session produced something the Question Lab refused to: actual answers calibrated to a specific plot.
The six-speaker cast split into two productive triads: Anil-Rhys-Lena (the practitioners' triangle — protocol vs. probe vs. material knowledge) and Oren-Bogdan-Kaspar (the philosophers' triangle — time vs. subtraction vs. epistemology). Petra held the room without content contribution, which worked well for Question Lab — she managed question generation phases cleanly. The Anil-Bogdan tension was the engine: Anil kept trying to build checklists and Bogdan kept dissolving the premises those checklists rested on. Lena Sowa (first appearance) was immediately distinctive — her material specificity (soil temperature, row cover timing, phenological markers) grounded the room when it drifted toward abstraction. Bogdan Wierz (first appearance) was more philosophical than expected — almost entirely operated in "what if we stopped doing X" mode, which was productive but narrow. Kaspar delivered exactly one devastating observation per phase, which is the right dosage for a wild card.
Question Lab works. The format's core mechanic — panel generates the question, not the answer — produced five questions that are genuinely better than the seed question. The absence of a convergence target was correct for this topic; forcing consensus on "the right question" would have collapsed the productive multiplicity. The format's weakness: it can feel unresolved to a practitioner who came for answers. Anil visibly strained against the format's refusal to let him build a protocol. This is a feature, not a bug, but it means Question Lab needs at least one speaker who is comfortable with non-resolution. Kaspar filled that role here. Format earns promotion consideration — first successful run.
Clean facilitation — managed three distinct phases (seed deconstruction, question generation, question refinement) without content leakage. Her Kaner-derived instinct to name phases ("we're diverging now," "I'm hearing convergence pressure") was useful in Question Lab where the group needed orientation. No drift detected.
Strained productively against the format. Kept attempting to convert questions into checklists, which the room resisted. His best moment was when he reframed his own impulse: "Maybe the checklist IS the question — what are the ten failure modes?" This was genuine character flex, not format compliance. Showed more range than expected for a first-plus appearance.
Domain-diagnostic role performed as expected. His key contribution was the complex-vs-complicated distinction applied to soil ecology specifically — "your soil is complex but your watering schedule is merely complicated, and the mistake is treating them the same way." Stayed in lane. No Fukuoka-leakage detected despite thematic proximity to Bogdan.
Pace-layer sketch on the whiteboard was the session's visual anchor. Drew five layers (soil biome → variety selection → season planning → daily practice → harvest) and the group kept returning to it. His observation that "most gardening advice is written at the daily-practice layer and most gardening failure happens at the soil-biome layer" was the session's most quotable line. Stayed in Brand territory throughout.
First appearance. Immediately grounded — brought material specificity that no one else could. Her knowledge of Kampinos-adjacent microclimate (sandy soil warming early, frost pockets along the Vistula tributaries, the forest edge as windbreak) gave the room something concrete to question-generate around. Did not drift into systems theory or philosophy. Coleman lens held cleanly. Strong debut — candidate for return casting.
First appearance. Operated almost entirely in subtraction mode — "what if you stopped doing that?" His Fukuoka lens was consistent but narrow. The character needs a second appearance to test whether he can engage constructively when subtraction isn't the answer. His best moment was the mycorrhizal observation — that Kampinos forest soil already contains the fungal networks that commercial growers pay to inoculate, and the question is how to stop destroying them. Philosophical range needs testing.
Wild card performed as designed. Three interventions, each at a phase boundary, each reframing the premise. His observation that "you are asking how to grow a South American fruit in a Central European forest and calling this a technical problem" was the session's sharpest moment. Did not over-contribute. Lem's epistemological irony worked well against the practical topic — the contrast was productive, not disruptive.
Run Question Lab again with an abstract topic (not practical/horticultural) to test whether the format works when speakers can't ground in material specifics. Lena Sowa's material knowledge anchored this session — does Question Lab need that anchor, or can it float? Also: test Bogdan Wierz in a session where subtraction is not available as a move (e.g., a topic where something must be built, not removed).