EXP-04 · Trade·Siting instrument
Hub Map
Hemp scales hub by hub, not field by field. A decortication plant can only buy stalk from a ring around it. The Hub Map shows, input by input, where supply, offtake and legal clarity overlap well enough for a lender to fund one. It publishes the arithmetic, never a verdict.
MARKET 1.6×SCHEMATIC BOUNDARY · 1 SQUARE = 50 KMGenerated fields
01The instrument
Ten candidate sites, one ring, your weights
Pick a site on the map or in the ranking. Move the ring, the intake and the weights: the map, the score, the ranking, the curve and the siting report below all recompute.
◎Grand Est · schematic
◎Candidate site
Marne
Châlons-en-Champagne
medium confidenceRing tonnage by evidence tier
What would change this score
- Raise documented-or-better supply from 38% to 45%: match declared fields to parcel records and sign contracts for 5.1k t more.
- Capturable stalk covers 98% of intake. Either plan a smaller line (class B) or recruit about 81 ha of new planting inside the ring.
- 59% of ring stalk goes to Bar-sur-Aube and Seine-et-Marne. A supply agreement with them changes this more than any new field.
- Known tender demand absorbs 31% of hurd output. Name one private offtaker (board maker, block maker) before lending.
◌Ring and plant
Plants buy from a 75 to 100 mile radius (New Frontier Data, from a summary, to verify). Bulky stalk does not travel.
Dry stalk per year. A lender sets this from the machine and the business plan.
⚖Weights · the lender's, not ours
≡Candidate sites, ranked
Bar colour = confidence · L low · M medium · H high
∿How the score moves with the ring
02Who opens it, and what they ask
Four people stand around one plant
The deal for a processing hub needs a lender, a region, a machine and growers to say yes at once. Each comes with a different question; each answer is the same record seen from another side.
●The lender · funds
“If I fund a line here, is there enough stalk under contract, and who buys the hurd?”
- Opens
- A siting report for one candidate, with every input sourced and tiered.
- Needs
- Supply cover at a realistic yield, rival plants in the ring, named offtake, legal status at both ends, and what would make the score wrong.
- Leaves with
- A due-diligence appendix for a credit committee, and a list of evidence to demand from the sponsor.
●The region · co-funds
“Which of our towns should host the plant, and what does it do for farm income?”
- Opens
- The ranked list across its territory, plus the corridor numbers.
- Needs
- Comparable sites under the same weights, and the public tenders it already runs that the plant could supply.
- Leaves with
- A shortlist, and a programme to run the corridor.
●The plant builder · supplies the machine
“Which machine class fits this ring, and who already runs one nearby?”
- Opens
- The hardware registry from the site card: classes that fit the capturable tonnage, with open designs and measured installs.
- Needs
- Feedstock profile (retted or green, stalk length), power and space at the site.
- Leaves with
- A qualified lead, and an install record once the plant runs.
●The grower · decides to plant
“If I put 40 ha into hemp, is there a plant close enough to buy my bales?”
- Opens
- The same map from the farm page, centred on their own fields.
- Needs
- Distance to the nearest buyer and to any planned hub, and how firm that plan is.
- Leaves with
- A reason to sign a supply contract, the strongest input in the lender's report.
03How a site is scored
Four plain ratios, weighted by whoever carries the risk
Deliberately simple: a lender should be able to rebuild the score on paper from the appendix. The inputs and the arithmetic are published; a verdict like “build here” never is, only what clears under stated weights.
ƒThe arithmetic
capturable = Σ fields in ring ( area × yield × your share of the field's pull × border factor ) supply = min(1, capturable / intake target) offtake = min(1, tender demand within 1.6 × ring / (intake × hurd share)) free = 1 − stalk lost to rival plants / stalk in ring (a field splits between plants in range by 1 / distance^1.5) evidence = min(1, share of ring tonnage at documented or verified tier / 0.5) score = 100 × Σ weight × ratio / Σ weight · confidence = low under 20% documented-or-better, high from 45% assumptions shown: yield 6.5 t dry stalk/ha · hurd share 0.62 · border factor 0.85 for stalk across a national border
◆A score never travels without its inputs
Every number on the map opens to its source, tier and last-checked date. A screenshot of a score with no appendix is the failure this exists to prevent.
◆Declared is not observed
Field area starts as declared (parcel registers, grower claims). Satellite classification later turns it into observed. The two are shown side by side, never merged.
◆Low confidence asks for evidence, not money
A high score with low confidence produces an evidence list (contracts to sign, fields to verify), not a recommendation to lend.
◆No one pays to move a site
A report buys the document. It never buys a weight, a rank or a hidden rival. Weights are the reader's and stay visible on every export.
≡Inputs, sources and the tier each can reach
| Input | Where it comes from | State today | Tier it can reach | Refresh |
|---|---|---|---|---|
| Declared hemp area by parcel | France: the annual RPG parcel register (open data). Elsewhere: national IACS/LPIS extracts where published | ○ not held | Declared | Yearly |
| Observed hemp area | Sentinel-2 crop classification with a research partner (see Eyes in the Sky) | ○ not held | Documented | Per season |
| Supply contracts | Growers and processors attach them on their claimed pages; the counterparty is checked | ○ not held | Documented to verified | On change |
| Existing processors and capacity | Company records (430 public records, 43 countries) and research classifications; capacity as claimed | ◐ partial | Declared to documented | Quarterly |
| Offtake demand | Tender feed (107 EU notices mentioning hemp since Jan 2024), plus demand-desk requests | ◐ partial | Documented | Daily |
| Legal status, origin and destination | Legality Atlas (237 territories, 1,820 current facts) | ● live | Documented | Per fact cadence |
| Value-chain fit | Origin stages and edges: which outputs (fibre, hurd, dust) feed which products | ● live | Reference | On change |
| Machines and installs | Open processing hardware registry (below) | ○ not held | Declared to verified | On install |
| Roads, power, zoning | Open map data and site visits; share-alike licences recorded | ○ not held | Documented | Per report |
04What the lender reads
The siting report
One candidate, one document, rebuilt from the map with the reader's weights. This one is generated live from the site selected above: pick another site or move a slider and it rewrites itself.
Châlons-en-Champagne: 67 of 100, medium confidence
Does not clear under the stated weights. The binding constraint is offtake.
1 · Stalk supply
71.6k t of dry stalk a year grows inside the 75-mile ring at an assumed 6.5 t/ha. After sharing with rival plants and discounting 684 t of cross-border stalk by 15%, 29.5k t is capturable, which covers 98% of the intake target.
| Tier | Tonnes / yr | Share | Source in the real report |
|---|---|---|---|
| Declared | 44,433 | 62% | Parcel register crop declarations, grower claims |
| Documented | 24,185 | 34% | Signed supply contracts, satellite-observed area |
| Verified | 2,948 | 4% | Delivered tonnage at a processor scale, audited |
2 · Rival plants
Bar-sur-Aube (anchor processor (example)), Seine-et-Marne (out-of-region processor (example)) draw from the same ring. 59% of ring stalk is expected to go to a rival, splitting each field between plants in range by inverse distance. A real report replaces that split with each plant's contracted volume where it is documented.
3 · Offtake
Public tender demand within market reach (120 miles, since boards and hurd travel further than bales) is 5.9k t a year, against 18.6k t of hurd output. That covers 31%. Tenders in reach: Paris region 2,600 t · Reims 700 t · Nancy 650 t · Metz 600 t · Troyes 400 t · Châlons 300 t · Épinal 200 t · Luxembourg 400 t.
4 · Legal and machine
Cultivation and processing of fibre hemp are legal in France under the EU 0.3% THC limit (current Atlas fact, documented tier). Recommended machine class: C · industrial, sized to 29.5k t a year.
5 · What would change our mind
- Raise documented-or-better supply from 38% to 45%: match declared fields to parcel records and sign contracts for 5.1k t more.
- Capturable stalk covers 98% of intake. Either plan a smaller line (class B) or recruit about 81 ha of new planting inside the ring.
- 59% of ring stalk goes to Bar-sur-Aube and Seine-et-Marne. A supply agreement with them changes this more than any new field.
- Known tender demand absorbs 31% of hurd output. Name one private offtaker (board maker, block maker) before lending.
6 · Assumptions and method
- Yield 6.5 t dry stalk/ha, hurd share 0.62, border factor 0.85. All three are parameters the reader can change.
- Score = weighted mean of four ratios, published in full. The weights are the reader's; no verdict is set for them.
- Declared and observed area are reported separately and never merged.
05The other half · open processing hardware
A map tells you where. The machine decides whether.
Decortication is the binding constraint in fibre hemp, and the first plant in a new region carries the whole cost of the machine. The registry hosts open designs in one comparable format and tracks every install as a measured record, so each install appears on the map with real throughput instead of a brochure figure.
·Class A · field edge
Mobile scutcher
Runs at a farm cooperative, turns bales into baled fibre and loose hurd so the bulky part travels less.
- Fits
- Capturable under 6,000 t a year, or the first season of a new region
- Feedstock
- Field-retted, baled
- Outputs
- Total fibre and hurd, minimal cleaning
- Sited at
- A cooperative yard; needs a concrete pad and three-phase power
- Openness
- Target: full design files and bill of materials under an open hardware licence
·Class B · regional line
Modular decortication line
The hub most regions actually need: a fixed line built from standard modules so it can grow with supply.
- Fits
- Capturable from 6,000 to 25,000 t a year
- Feedstock
- Retted or green stalk; straw handling to ASTM D8513-24 guidance
- Outputs
- Fibre, graded hurd for hempcrete, dust for boards or fuel
- Sited at
- An industrial plot near the ring's centre of mass, road access for bale trucks
- Openness
- Modules documented to a shared interface so makers can supply parts of one line
▶Class C · industrial
Licensed industrial plant
Proprietary lines from established makers, listed in the registry on the same terms, with measured installs.
- Fits
- Capturable above 25,000 t a year with signed contracts
- Feedstock
- Year-round bale storage required
- Outputs
- Technical fibre to spec (e.g. GOST R 58957-2020 for undifferentiated fibre), hurd grades, boards
- Sited at
- Rail or barge access, storage for a season of bales
- Openness
- Closed design; open performance data is the condition for a listing
▤A design record, as the registry would show it
[Example design] Two-module breaker and cleaner, v0.4
Every field carries its own tier, the same three tiers the Hemp Passport uses. A measured throughput comes from an install record signed by the operator and checked against delivered tonnage, never from the maker's sheet.
↻How the two halves feed each other
- The map finds a site whose ring clears for a class B line.
- The site card lists open and licensed designs that fit the tonnage and feedstock, with their measured installs.
- The lender's report carries the machine's measured throughput, the number credit committees distrust most today.
- The plant is built. Its install record becomes a processor pin with real capacity, and its ring is subtracted from every rival candidate.
- Growers inside the new ring see a buyer and sign contracts; the region's evidence tier rises. The next report gets more confident.
- The tender feed also catches machinery calls, such as a grant-funded request for quotes for a hemp decorticator, and routes them to registry makers.
06How it gets built
One region by hand before one line of map code
The first report is made by hand from data that can already be reached, and shown to one lender. If nobody wants the document, the layer is not worth building.
Hand-built Grand Est report
Parcel register, processors, tender feed, Atlas. Three candidate sites, one lender or regional bank as reader.
Done when a lender reads it and asks for a second site or region.
Hub Map layer
Fields, processors, tenders and rings on the globe, with the site card and report export.
Done when a Grand Est report is generated from the map, not by hand.
Observed beside declared
Observed acreage from the satellite pilot beside declared. The hardware registry opens with the first makers and installs.
Done when one install's measured throughput appears in a report.
Reports and a data licence
Siting reports and a data licence for lenders, a second corridor, the growers' view on the farm page.
Done when a hub is financed with this report in its file.
✕Kill test
If three hand-built reports reach lenders and none asks for another by December 2027, the layer stops and the work folds into corridor programmes instead.
∅Out of scope
No machine building, no equity in plants, no offtake brokerage beyond the demand desk. The Hub Map stays the record every party can check.
07Open decisions
Five calls this experiment still needs
?01 · Who reads v0?
One lender, regional bank or bio-based industries consortium in the Grand Est orbit who would read a hand-built report.
?02 · Are default weights ours at all?
This alpha ships with lender-style defaults. The alternative is no defaults, forcing every reader to choose.
?03 · Do contracts go on the record?
Supply contracts are the strongest input, and commercially sensitive. They could stay private and appear only as counted tonnage at a tier.
?04 · How open is the hardware registry?
Open designs only, or licensed industrial lines too, on condition of open performance data.
?05 · Does the public layer show candidate sites?
Showing them may move land prices and local politics. The safer default shows fields, plants and tenders publicly and keeps candidates inside reports.