Feral Index

The Feral Index measures how ready each basic human need is to be freed from labor through automation — using only technologies that cannot be monopolized or suppressed.

Data as of 2020

Liberation of human needs

Each card is a basic need. The score shows how ready it is for automation with simple, accessible technology. Data: Eurostat 2024 household expenditure (COICOP), OECD ICIO 2020.

62% of household spending goes to needs that have been automatable with pre-1900 mechanical technology. The technology exists. The question is who controls it.

Industry explorer

Detailed industry-level data behind the needs analysis. Each bubble is an industry sector from the OECD ICIO classification.

Industry map

Each bubble is an industry. Position: autonomy (X) vs leverage (Y). Size reflects labor intensity. Color reflects sector group. Click a bubble to see details.

Automation cascade

Optimal automation order based on supply chain depth. Wave 1 industries have the simplest inputs and can be automated first. Each subsequent wave depends on the previous ones.

Wave 1
Wave 2
Wave 3
Wave 4
Support infrastructureUniversal intermediaries needed by all sectors. Can be automated independently at any stage.

Industry rankings

SectorFeral ScoreLeverageAutonomyAccessibility
Textiles, clothing, leather
100
88
82
100
Agriculture, forestry
100
40
91
100
Wood products
100
76
92
94
Non-metallic minerals
97
72
71
94
Paper, printing
96
76
75
88
Food, beverages, tobacco
95
77
100
94
Mining (non-energy)
93
47
63
94
Electricity, gas supply
92
56
70
81
Fishing, aquaculture
91
38
85
75
Water, waste management
90
45
66
88
Warehousing, support
90
45
68
88
Fabricated metals
90
83
51
81
Other manufacturing
89
67
61
69
Accommodation, food services
89
46
87
63
Mining (energy)
88
35
69
94
Mining support
88
55
67
69
Rubber, plastics
88
83
55
88
Chemical products
84
80
52
75
Land transport
83
44
60
69
Water transport
83
74
61
63
Postal, courier
80
34
67
69
Machinery, equipment
77
82
37
63
Electrical equipment
74
94
35
69
Air transport
73
67
61
38
Pharmaceuticals
72
51
45
50
Basic metals
71
99
61
94
Wholesale, retail trade
70
25
66
56
Construction
70
69
64
25
Motor vehicles
70
100
29
88
Other transport equipment
67
78
24
56
Other services
66
21
67
50
Administrative services
57
16
61
44
Telecommunications
55
33
34
31
Finance, insurance
52
17
48
38
Publishing, audiovisual
52
32
48
19
Public administration
51
18
60
25
Health, social work
42
19
54
13
Professional services
40
31
50
6
Arts, entertainment
40
22
73
6
Real estate
14
0
66
13
Petroleum, coal products
0
82
81
75
Electronics, optical
0
79
0
31
IT, computer services
0
24
39
0
Education
0
0
70
0
Methodology

The Feral Index combines four dimensions to identify which industries, once automated, would maximally liberate human labor while remaining resilient to suppression by capital interests controlling advanced technology supply chains.

Data source

We use the OECD Inter-Country Input-Output (ICIO) tables (2020), which record monetary flows between 45 industry sectors across 80+ economies. The table is aggregated into a single world input-output matrix. Universal intermediary sectors (wholesale trade, finance, professional services, admin, real estate) are separated in cascade analysis because they appear in every industry's inputs as transaction infrastructure, not as physical production dependencies.

How leverage is computed

From the world IO table, we compute the technical coefficient matrix A (where a_ij = flow from i to j / total output of j), then the Leontief inverse L = (I - A)^{-1}. The leverage of sector j is the column sum of L — the total output generated across the entire economy per unit of final demand for j. This captures both direct and indirect cascade effects.

How autonomy is computed

For each sector, we look at its input column in the technical coefficient matrix and compute the weighted average complexity of its inputs. Complexity is assigned per ISIC section and refined for manufacturing sub-sectors based on the OECD R&D intensity taxonomy and economic complexity literature. Autonomy = 1 minus this weighted complexity. Sectors that consume mostly primary resources (agriculture, basic minerals) score high; sectors that depend on electronics or pharmaceuticals score low.

How automation accessibility is assessed

Automation accessibility measures what minimum technology level is needed to meaningfully automate a sector (~30% of tasks). Four levels: (1) Mechanical — gears, steam, water power, zero chip dependency; (2) Electromechanical — motors, relays, conveyors; (3) Basic electronics — PLCs, 8-bit microcontrollers, producible on >180nm nodes without EUV lithography; (4) Advanced — modern CPUs, AI, computer vision, requiring <7nm chips from TSMC/ASML. Scores are based on historical evidence (many sectors were automated in the Industrial Revolution), McKinsey's 18-capability framework, and chip fabrication independence analysis. Sectors scoring >= 0.65 are automatable without any dependency on advanced semiconductor supply chains.

How labor intensity is estimated

Labor intensity is measured as the value-added share of gross output, computed directly from the OECD ICIO table: VA = total output minus intermediate consumption. Higher VA share means more of the sector's output goes to labor compensation and operating surplus rather than buying intermediate inputs. This is a standard IO-analysis proxy for labor intensity — sectors with high VA share employ more labor per unit of output. Services score high (education ~76%, public admin ~64%); capital-intensive manufacturing scores low (petroleum ~22%, basic metals ~23%). Future versions will incorporate OECD TiVA LABR (labor compensation) data for a more precise decomposition.

How the Feral Score is computed

The Feral Score is a weighted geometric mean of four normalized axes: leverage (weight 2.0), autonomy (1.5), accessibility (1.5), and labor intensity (1.0). Leverage is weighted highest because cascade effect is the primary strategic value. Autonomy and accessibility are equal — both measure resistance to suppression, but through different mechanisms (input supply chains vs. automation technology supply chains). Labor intensity is the lowest weight as a liberation bonus. The geometric mean ensures a sector must score well across all dimensions — a single zero collapses the entire score.

The bootstrap strategy

The automation cascade shows the optimal order: start with sectors that have the simplest inputs AND can be automated with the simplest technology (Wave 1: textiles, food, wood, agriculture). Use the freed capacity to build basic electronics manufacturing (180nm+ fabs, ~$100M vs $10B+ for advanced fabs). Use basic electronics to automate Wave 2-3 more efficiently. This is not a confrontation with capital — it is building a parallel infrastructure from the ground up, using resources and technologies that no one can monopolize.

Limitations

The current version uses monetary IO tables (not physical flows), sector-level complexity estimates (not product-level PCI mapping), value-added share as a labor intensity proxy (not direct employment counts from OECD STAN/TiVA LABR), and expert-assessed automation accessibility (not empirically measured). Geographic concentration of inputs is not yet factored in. The 44-sector OECD classification is coarse — real-world automation decisions require sub-sector granularity.

Automation accessibility: per-sector justification

Each sector's accessibility score is based on the minimum technology level needed to automate ~30% of tasks. Scores and justifications are derived from historical evidence, McKinsey's 18-capability framework, and chip fabrication independence analysis.

SectorScoreMin levelJustification
Agriculture, forestry901-2Mechanical threshers, seed drills, steam plows displaced 30%+ labor in 1800s; electromechanical irrigation adds more
Textiles, clothing, leather901Canonical pre-electronic case: power looms (1784), spinning frames automated dominant activities
Mining (energy)851-2Steam-powered hoists, pumps (Newcomen 1712), mechanical conveyors automated most dangerous tasks pre-electronics
Mining (non-energy)851-2Crushers, grinding mills, conveyors, hoists all mechanical; 19th-century mechanized hardrock mining
Food, beverages, tobacco851-2Continuous-flow canning, bottling, grain milling mechanical since 1800s; conveyor-and-fill needs no PLCs
Wood products851-2Water-powered sawmills (1500s), mechanical planers, band saws; electromechanical log sorting
Non-metallic minerals851-2Rotary cement kilns, glass-blowing machines (1903), tile extrusion presses -- all mechanical
Basic metals851-2Blast furnaces, rolling mills, continuous casting -- mechanized in 1800s-1950s before programmable logic
Paper, printing802Fourdrinier papermaking machine (1806, mechanical); printing presses mechanical since Gutenberg
Rubber, plastics802Injection molding, extrusion, vulcanization presses are electromechanical; relay-based cycle timers
Motor vehicles802-3Body stamping/welding automated 1950s (Ford transfer lines, relay logic); full assembly needs PLCs
Water, waste management802-3Pumping stations, filtration, conveyor-based waste sorting electromechanical; relay-controlled pump cycling
Warehousing, support802Conveyors (1901), electromechanical AS/RS (1960s Demag), carousel systems -- zero advanced chips
Fabricated metals752-3Stamping presses, welding rigs mechanical; PLCs needed for sequencing and quality interlocks
Electricity, gas supply752-3Turbine governors, boiler controls used mechanical/pneumatic pre-1970; substation relays electromechanical
Fishing, aquaculture702Powered winches, mechanical fish processing (gutting/filleting); sonar not needed for 30% automation
Petroleum, coal products702-3Pneumatic analog instrumentation achieved substantial process control in 1950s-60s refineries
Chemical products702-3Batch reactors, pumps, mixing via pneumatic analog; safety margins push to PLCs
Mining support652-3Drilling rigs are mechanical but safety interlocks and blasting timers need PLCs
Electrical equipment653Transformer winding, motor coil insertion have electromechanical paths; wire routing needs PLCs
Other manufacturing653Heterogeneous: furniture cutting mechanical, but medical devices need precision/traceability via PLCs
Land transport652-3Rail signaling electromechanical since 1930s; road freight loading uses conveyors; driving itself needs L4
Postal, courier652-3Mechanical sorting conveyors Level 2; address-based sorting (OCR for zip codes) needs Level 3
Machinery, equipment603CNC (Level 3) for metal removal; variety of parts resists pure mechanical automation
Water transport602-3Port cranes, conveyor loading, gyroscopic autopilot (1920s) are Level 2; 30% achievable
Accommodation, food services602-3Dishwashers, conveyor grills, dough mixers, HVAC electromechanical; customer-facing resists automation
Other transport equipment553Large, low-volume, customized; structural fabrication automatable but variety needs programmable control
Wholesale, retail trade552-3Conveyors, automatic sorters, mechanical carousels handle 30%+ warehouse work; retail resists low-tech
Pharmaceuticals503FDA validation mandates documented computer systems; precision dosing requires electronic sensing
Other services502-3Industrial laundry, automated dry-cleaning, vehicle washes electromechanical; repair/personal care manual
Administrative services453Data entry, scheduling, call routing, document sorting automatable with rule-based systems since 1970s
Air transport403-4Flight management needs precise sensors/real-time computing; ground handling electromechanical
Finance, insurance403-4Back-office transaction processing since early computers; 30% sector-wide needs algorithmic trading/fraud detection
Electronics, optical353-4Making electronics requires electronics: pick-and-place, AOI, reflow ovens need Level 4 precision
Telecommunications353-4Strowger crossbar switching (1920s) was electromechanical; modern packet routing needs Level 4
Construction304Unstructured, changing sites; concrete mixing/hoisting automatable but site navigation needs computer vision
Public administration303-4Form processing/records at Level 3; governance/adjudication/citizen services human-judgment-intensive
Publishing, audiovisual254Printing presses mechanical, but dominant labor is content creation/editing -- needs Level 4 AI
Real estate204Valuation, negotiation, tenant management are human judgment; building management is small fraction
Health, social work204Hospital logistics (sterilization conveyors, dispensers) electromechanical; patient care irreducibly human at L1-3
Professional services154Legal, accounting, engineering, R&D -- expert judgment and creative problem-solving; McKinsey ~35% with best tech
Arts, entertainment154Live performance, creative production, participatory sports fundamentally human; tech support is small fraction
IT, computer services104The work product IS advanced computing; no Level 1-3 path to 30% automation
Education104Teaching, curriculum, mentoring are core work; McKinsey estimated only 27% with current best tech
Sources & references