A Tomato Is a Map of Power

I took a photograph of a bag of tomatoes because the origin looked unfamiliar.

The package was not trying to hide. It said Dakhla Oued Ed-Dahab in small black letters. It also gave me a lot number, a packing-station code and a GLOBALG.A.P. number. This was more traceability information than I normally receive from a tomato, but I still did not know where it was from.

I asked the obvious question first: where is Dakhla Oued Ed-Dahab?

The answer opened a trapdoor. The tomatoes came from the part of Western Sahara administered by Morocco. Western Sahara is still on the United Nations list of non-self-governing territories. Morocco controls most of the territory, the Polisario Front seeks an independent Sahrawi state, and the referendum that gave the UN mission in the territory its name has never happened.

Then the questions kept changing.

Is there a war? Can you fly there on Ryanair? Which company grows tomatoes in the desert? Where does the water come from? Who receives the money when I buy them in Estonia? Does any of it reach Sahrawi refugees in Algeria? At what point does a packet of tomatoes stop being a food product and become a political, environmental and financial system?

This is the kind of chain I have been trying to build into Green Filter. The original idea was a shopping companion that could help someone understand whether a product is sustainable. The tomato made the limitation of that sentence obvious. Sustainability is not one property waiting inside the barcode. It is a series of questions that appear only after the first answer changes what the user needs to know.

The package contained data, but not an explanation

The visible facts were unusually specific:

• Rimi mini plum tomato mix
• One kilogram
• Origin: Dakhla Oued Ed-Dahab
• Packing station: 7607R
• GGN: 4049929293588
• Lot: 1500.39827461306

Rimi's online shop currently lists the product at €6.99 per kilogram, although the origin on a retailer page can change between shipments. The physical package matters more than the generic product page because it is the only record tied to this particular bag.

At first, the codes made me think the grower would be easy to identify. They were not. The GGN did not lead to a public farm record I could verify. The station code appears on Carrefour tomato listings in Belgium as well, which suggests that it identifies a packing operation used across retailers rather than the greenhouse where these tomatoes grew.

Several companies operate tomato farms around Dakhla. Azura is one of the largest and its Maraissa entity runs eight tomato farms there, but I could not connect this exact lot to Azura with enough confidence to call it the supplier. That distinction matters. The strongest available match is not the same thing as identification.

A normal product database is uncomfortable with that answer. It wants a row containing a brand, supplier, country and score. A useful Green Filter result should be allowed to say:

Exact farm: unknown. Major regional producers are identifiable, but this package does not provide enough accessible evidence to connect the lot to one of them.

The absence is not empty. It tells us that the supply chain is traceable to someone inside the system, but not understandable to the person paying for it.

A region name can hide a territory

Dakhla Oued Ed-Dahab sounds like an ordinary administrative region. It does not tell an Estonian shopper that the tomato was grown in Western Sahara.

That wording has a strange legal history. In October 2024, the Court of Justice of the European Union ruled that tomatoes and melons harvested in Western Sahara had to identify Western Sahara alone as their origin. The Court treated the territory as separate from Morocco and said another indication could mislead consumers.

One year later, a new EU-Morocco arrangement began applying provisionally. The accompanying EU regulation replaced the country indication for these products with the name of the production region, specifically Dakhla Oued Ed-Dahab or Laâyoune-Sakia El Hamra. The words on this package are therefore not a random euphemism invented by Rimi. They are part of the current European trade and labelling system.

The system technically distinguishes the produce from Morocco, but it does so in language that most shoppers cannot decode. The label is precise and opaque at the same time.

This may be one of the most useful Green Filter categories: information that is legally present but practically absent.

The packet tells me the region. Green Filter would need to explain that the region is in a disputed, non-self-governing territory administered by Morocco. It would also need to avoid turning that sentence into a fake geopolitical verdict. Morocco describes the territory as part of the country and proposes autonomy under Moroccan sovereignty. Polisario and the Sahrawi Arab Democratic Republic claim independence. The UN process remains unresolved.

The consumer does not need the entire history of the conflict before making dinner, but they should not have to know the history already in order to understand the label.

The water question changed the environmental story

The tomatoes were grown in one of the driest landscapes connected to the European supermarket system. The first environmental question was not packaging or food miles. It was water.

The most detailed public evidence I found is an Alliance for Water Stewardship audit of Azura's Maraissa operation in Dakhla. Again, this does not prove that Azura grew this exact bag. It does show how a major tomato producer in the same regional system was using water.

The audit covered eight farms managed as one site. It said every farm used borehole water and none used desalinated water. The groundwater catchment was the Sahara Aquifer, whose extent and boundaries were not well mapped. The water had high salt content, one farm needed to remove salt, and part of another farm sat inside the Dakhla Bay Ramsar wetland area.

The report also showed how much remained unfinished. The site did not yet have a complete water balance. It was not measuring every outflow, had not accounted properly for losses, was not tracking withdrawals against authorisation limits for each borehole, and did not yet have an annual target for irrigation water-use efficiency. Azura had asked to buy water from a proposed desalination plant in order to reduce borehole withdrawals, but the audit still found the farms entirely dependent on groundwater in 2024.

The operation nevertheless received AWS Core certification, valid until May 2027. The audit raised one major non-conformity, twenty minor non-conformities and two observations. The major issue was closed and corrective plans were accepted.

That is not evidence that certification is meaningless. It is evidence that a certification badge compresses a process.

A shopper sees a logo and reasonably assumes that someone checked the problem. Someone did. The full report then reveals a much more useful picture: borehole dependence, saline water, incomplete measurement, planned corrections and a producer trying to move toward a different water system.

Green Filter should not translate that into either certified, therefore good or desert groundwater, therefore bad. It should show the structure:

• Water source: boreholes
• Desalinated irrigation water: no, at the time of the audit
• Aquifer condition: insufficiently mapped
• Product-level litres per kilogram: unavailable
• Water-efficiency target: not yet established at the time of the audit
• Wastewater treatment and reuse: present
• Certification: AWS Core
• Confidence: high for the audited Azura farms, unknown for this exact packet

The environmental result is not a score. It is a set of claims with different scopes.

The money does not travel to one place

The next question was the one Green Filter should probably place closer to the beginning: who does my money support?

The €6.99 shelf price does not go to a farm in Dakhla. It separates into retailer revenue, Estonian and Baltic distribution, refrigerated transport, packaging, packing, rejected produce, importer and exporter costs, farm operations, wages, energy, water infrastructure, taxes and margins at several points.

The exact split is not public. The political direction is easier to see than the accounting.

Buying the tomatoes supports Rimi and the companies moving the product through Europe. It supports whichever grower and exporter supplied this lot. It pays some wages to farm, packing and logistics workers. It also contributes to the Moroccan-administered commercial economy that has been built around Dakhla.

It does not create an identifiable payment to the Polisario Front, the Sahrawi Arab Democratic Republic or the refugee camps near Tindouf in Algeria. There is no mechanism on the packet through which part of the price is transferred to Sahrawi representative institutions.

This does not mean nobody Sahrawi can benefit from the farms. Some workers or residents may be Sahrawi, and regional employment is real. The European Commission's 2025 proposal cited Moroccan data showing 74,000 tonnes of agricultural products from Western Sahara exported to the EU in 2022. It also estimated that the broader trade arrangement supported more than 49,000 direct jobs across the territory and saved importers €44.4 million in duties in 2023.

Those figures describe economic activity among the people currently living in the territory. The legal question asks something else: whether the people of Western Sahara, as the group holding the right to self-determination, consented to the use of the territory and received a specific collective benefit.

The EU Court said the previous agreement did not establish that consent. It also found that general economic advantages were not the same as a proportionate, verifiable benefit provided to the Sahrawi people in exchange for the exploitation of resources.

Both statements can be true. The industry can create jobs inside Moroccan-administered Western Sahara while the collective political and financial benefit to the Sahrawi people remains unproven.

A one-dimensional ethical label cannot hold that contradiction. Green Filter needs separate lines for employment benefit, ownership, collective consent, distribution of profit and confidence.

The Green Filter result I would want to see

After the research, the useful product card would not say sustainable or unsustainable.

It would say something closer to this:

Stated origin: Dakhla Oued Ed-Dahab Context: A Moroccan-administered region in the disputed territory of Western Sahara Exact farm: Not publicly verified from the package Water risk: Elevated. Major audited tomato farms in the region used borehole groundwater from a poorly mapped aquifer Certification: Relevant water certification exists for Azura's Dakhla operation, but this lot is not confirmed as Azura and the audit contained unresolved measurement and governance work Economic benefit: The trade supports regional employment and the Moroccan-administered economy Sahrawi consent and collective benefit: Not demonstrated by the available product or trade evidence Product carbon footprint: Not available Confidence: Mixed, with high confidence on territorial context and low confidence on the exact supplier

Then it should offer actions that match the evidence.

I could compare it with another tomato whose farm and water source are disclosed. I could decide that buying from a disputed territory without evidence of consent is a personal red line. I could still buy it because I value the jobs and do not think a supermarket boycott is a precise political tool. I could ask Rimi to identify the supplier, farm, water source and product-level footprint.

The important thing is that the interface should not pretend the decision is simple just because the user is standing in a shop.

Green Filter began as a way to reduce the amount of research required for sustainable shopping. This case suggests that the deeper role is not to remove complexity. It is to stop complexity from being hidden inside confident labels.

Traceability is not the same as accountability

The tomato packet had a lot number, a station number and a GGN. It looked like a product that could be followed backwards.

I followed it backwards and reached a point where the public trail stopped. The region was identifiable. The political status was identifiable. A major producer's water audit was identifiable. European trade volumes, court judgments and tariff arrangements were identifiable. The farm that grew the tomato was not.

This is a useful distinction for data-driven design. More data does not automatically produce more agency. The codes may be excellent for a recall, inspection or business-to-business audit while remaining almost useless for the consumer's actual questions.

Who owns the farm? Which borehole supplied it? How many litres of water were used per kilogram? Was the water extraction below the aquifer's recharge rate? What was the transport route to Estonia? How much of the price remained in Dakhla? Were Sahrawi workers represented in the audit? Who was consulted about the land and water? What happened when the certification found twenty minor problems?

A digital product passport could theoretically carry these answers. Most current traceability systems carry identifiers instead.

The Green Filter case is therefore not only about tomatoes or Western Sahara. It is about the difference between proving that a product moved through an authorised chain and explaining what the chain does.

A barcode can become a map of power

I began with an origin label and ended up with a map.

The map contains a Rimi shelf in Estonia, greenhouse farms in the Sahara, boreholes in a poorly mapped aquifer, a wetland, packing and logistics operations, European tariff preferences, a disputed border, a UN mission, Moroccan institutions, Sahrawi refugee camps in Algeria, workers whose individual backgrounds are not visible, and a consumer trying to understand what a purchase means before the tomatoes go soft.

That map is what Green Filter should reveal.

Not every product requires a geopolitical investigation. The point is that the interface should be capable of noticing when one does. Origin data should be expandable. Certifications should open into findings. Company claims should remain separate from product evidence. Unknowns should be shown as unknowns. Money should be treated as a flow through institutions, not only a price paid for an object.

Shopping is a small allocation of capital. The amount is tiny, but the direction is real.

A tomato is food. It is also water, land, labour, logistics, law and political recognition arranged into something that fits inside a plastic bag.

Connected reading

• The Receipt Remembers
• Every Interface Contains a Business Model
• The World Runs on Companies You Have Never Heard Of
• How Companies Actually Do Data-Driven Design

AI authorship note

This case study began with Kris Haamer photographing a bag of Rimi tomatoes and asking a sequence of increasingly specific questions about its origin, the conflict in Western Sahara, the grower, the water source and where the purchase money goes. AI supported source discovery, legal and supply-chain research, uncertainty tracking, drafting, editing and implementation. Kris directed the connection to Green Filter, kept the investigation centred on the physical packet, and chose to publish the result on Raw Drafts.