Nineteen Tables For One Business Model

13,561 bytes. 13,556 characters. One file, counted two ways, and the two counts differ by five. Those five bytes are the extra weight of the three characters in this document that sit above the alphabet: two em dashes at three bytes each and one enye at two, and two em dashes twice over plus one enye is five, so the gap closes to the byte. The highest code point in the file is 8,212, the em dash, and it appears twice. The average line runs 39.8 characters. The longest runs 344, on line 36. 332 lines on the raw split and 331 with the final newline, 89 of them blank and 243 carrying content, 2,231 words, 11,133 characters that are not whitespace, and not one line carrying a trailing space. The mode is 0600. This file is closed to its group and closed to the world, and it is a business model for a plant that does not exist yet. 31 heading lines sit over it: 1 title, 15 at the second level, 15 at the third, and nothing deeper than the third. 14 sections are numbered, 1 to 14, and not one number is missing or out of place. That is the only count in the document that agrees with its own list. The structure below the headings is a grid. 142 lines start with a pipe, which counted as blocks gives 21, and counted as tables gives 19. The difference sits inside the one fenced block in the file, lines 197 to 209, where two single pipes on lines 199 and 201 are the vertical strokes of an organisation chart. Inside a fence a pipe is a stroke. Outside a fence it is a column. Two lines change meaning on whether a reader counts fences, and a parser that does not count fences reads 21 tables where the author drew 19.

The 19 tables take 140 lines: 19 header rows, 19 separator rows and 102 rows of data. That is 42.3 per cent of the 331 lines in the document. The widest table has 5 columns and there is exactly one of them. 3 columns appear in 8 tables, 2 columns in 7, 4 columns in 3. Not one row in any table is ragged: every row carries the same number of cells as its header, in all 19. The narrowest table runs 3 rows of data, the widest runs 11 and it is the last one in the file, the steps table that closes the document. Money runs through the grid rather than beside it: 49 of the 102 rows carry a EUR figure, 93 carry a number at all, and 8 carry a per cent sign. 268 numbers stand alone in the file and 60 of them are attached to EUR. Against that, the document holds 0 exclamation marks, 0 question marks, 0 links, 0 web addresses and 0 wikilinks. It holds 9 bullet lines and 17 rules. There is no address in it of any kind, so nothing in it can be checked by a reader who is not holding the file. The weight sits unevenly too. The heaviest second level section runs 187 words and the average of the 15 sits at 82. Four of the 15 carry no words of their own at all, because every word under them sits one level deeper. A reader who skims the second level of this document reads four empty headings over the economics of the business.

The table at line 174 has 5 columns, the widest grid in the file. Its fourth column is headed Number in Andalucia. The four rows under it read: about 200 cooperatives, about 40 large orujeras, about 30,000 hectares, about 50 processors. One column, four units of measure. A cooperative is a firm. An orujera is a firm. 30,000 hectares is a land area, not a customer. 50 processors is a plant count. The table below it, at line 183, has 4 columns and drops that column entirely, so the industrial half of the customer section never states how many customers it is aiming at. The province column carries the same wobble. Malaga is named 6 times and Sevilla 6, then Almeria 3, Granada 2, and Cadiz, Jaen, Cordoba and Huelva once each. The registered office reads Malaga or Sevilla province. The first phase names Malaga province, at Campanillas or Antequera. The next steps table, 200 lines later, reads Malaga or Cadiz province. Two different second options for one company, and the province is not decoration: the subsidy row states 60 to 70 per cent depending on province, so which province gets picked moves the money. A column that counts firms in one row and hectares in another cannot be summed by the reader who is being asked to act on it.

The carbon table at line 155 does its own arithmetic in the open and then does not use it. The propane row reads 600 tonnes a year. The grid row reads 1,904. The methane row sets out its formula in full: 1,825 tonnes of waste diverted from landfill, times 0.3 tonnes of methane per tonne of waste, times 28 for the warming potential, times a 30 per cent capture rate. Worked through, that is 4,599 tonnes. The row states 8,000 to 12,000. The total row states 10,500 to 14,500, and the first two rows plus the stated methane band give 10,504 to 14,504, so the total is consistent with the row above it and the row above it is not consistent with itself. Add the rows as they are written and the plant avoids 7,103 tonnes a year, which is 3,397 short of the band the document publishes. The 1,825 is the second number worth chasing. The gate fee row prices 2,190 tonnes a year at 80 EUR a tonne, which is 6 tonnes a day for 365 days. The methane row counts 1,825, and 1,825 is exactly five sixths of 2,190. The 365 tonnes that go missing are one tonne a day, and no line in the document says which day of the week the plant stops taking waste. Everything else in the unit economics closes to the unit: the four revenue rows add to 1,347,200 and the five cost rows add to 460,000, and the carbon revenue in the revenue table, 12,000 tonnes at 20 EUR, is 240,000 in the middle of the 210,000 to 290,000 band the methodology section states. The one figure that does not close is the one the whole investment case leans on.

13,561 bytes. 13,556 characters. The gap is 5.

The roadmap is four tables and they multiply the standard plant by a different multiplier each time. Phase 2 builds 3 more plants for 4 in total, and 4 times 3,025,000 is 12,100,000, which is exactly the 12.1M the investor row states. Capacity is exact too at 3.4 MWe and 7.2 MWth. Revenue is stated at 5.4M and comes to 5.39M. Net cash flow is stated at 3.5M and comes to 3.55M. The subsidy at 60 per cent is stated at 7.3M and comes to 7.26M. Every one of those is a rounding, and rounding is honest. Phase 3 is where it stops being a rounding. A further 10 plants take the fleet to 14, and 14 times 3,025,000 is 42,350,000 against a stated 42.4M, so 50,000 EUR appears in the fleet total that no plant in the fleet accounts for. Capacity is stated at 12 MWe against a computed 11.9, and at 25 MWth against a computed 25.2, so in one row the electrical figure is rounded up and the thermal figure is rounded down. Net cash flow is stated at 12.4M against a computed 12.42M and the subsidy at 25.4M against a computed 25.44M. The single plant table is the cleanest grid in the file and it reconciles: 850 kW at 8,000 hours at 0.10 EUR a kilowatt hour is 680,000, heat is 252,000, the gate fee is 175,200, carbon is 240,000, and the four add to 1,347,200 on the nose. O and M is 150,000 for the gasifiers, 170,000 for the engine, 60,000 for labour, 40,000 for transport and 40,000 for everything else, and that is 460,000 on the nose. The engine line is 0.025 EUR a kilowatt hour across 6,800 MWh, and it lands at 170,000 exactly. Payback is stated at 1.4 years and the arithmetic gives 1.36, which is the same number rounded. Then the return row. The internal rate of return is stated at 68 per cent, and the cash flow in the row above it, 1,210,000 out and 887,200 a year for 10 years, returns 73.0 per cent. On the full 3,025,000 with no subsidy it returns 26.5 per cent. The one return figure in the document matches neither end of its own model, and the payback beside it quietly uses the 60 per cent subsidy rate named after a single province while the capital table allows the rate to run to 70.

The vision section runs 40 words and every one of them is a number. 50 plants across Andalucia by 2035. A 45 MW fleet. 500,000 tonnes of waste diverted from landfill a year. 250,000 tonnes of CO2 avoided a year. Three of the four survive the rest of the document. 45 MW across 50 plants is 0.9 MWe each, which is the engine rounded to one decimal, and that closes. 500,000 tonnes across 50 plants is 10,000 tonnes a plant against the 2,190 tonnes the plant's own gate fee row takes in, so the vision asks for 4.57 times the throughput of the machine it is describing. 250,000 tonnes of CO2 across 50 plants is 5,000 tonnes a plant against the 10,500 to 14,500 the carbon methodology computes, so the vision sits at 0.48 to 0.34 of the number the same file derives. The waste is overcounted by a factor of four and a half and the carbon is undercounted by a factor of three, and both errors push the same document in opposite directions on the one page a funder reads first. The roadmap underneath it never arrives. Its highest plant total is 14, at Phase 3. Phase 4 begins in year 7 and names four countries and no plant count at all, so the document stops counting plants on the page where the plant count stops being small. 14 to 50 in the nine years to 2035 is 15.2 per cent compound a year, and no row carries that rate. The stamps at the top of the file are worth one more look. Created 2026-06-26 and modified 2026-06-26, the same day, and indexed 2026-09-23, 89 days later, which is the only date in the file that reflects reading rather than writing. The version line says August 2026, two months after the file's own first stamp. The foot of the document is dated August. The domain line ends with available, check, which is a task left standing inside the artifact. The group behind the plan has built gasification machines before, 200 of them across the UK and Europe, so this was never a document about whether the machine works. It is a document about whether the numbers in a document agree with the machine, and on the four counts that matter here they do not. The probability that the fifty plants survive contact with the plant's own meter is close to zero.