Part 7 of 8. Part 6 looked at security and built-in ways in.

What this post covers

  • Why the comparison is useful.
  • Seeds, and licence terms on living things.
  • Supermarkets, and the rules built for them.
  • Why technology has nowt like it.
  • What it tells you.

Why the comparison is useful

There is a common view that technology is a new kind of power and needs a new kind of thinking.

It is a comforting view, and it is mostly wrong. Believing it is one reason the answer has been so slow coming.

The problem in this series is an old one. A supplier becomes essential. The customer’s alternatives disappear. Terms stop being agreed and start being announced.

That has happened in industries with no software in them at all.

The comparison helps 2 ways.

It shows what tends to come next, because those industries are further down the road.

And it shows what a serious answer looks like, because some of those answers already exist and work.

Seeds, and licence terms on living things

A farmer buying seed is in much the same spot as an organisation buying core software.

Four companies, Bayer, Corteva, Syngenta and BASF, hold most of the global seed market and a similar share of pesticides. The Swiss organisation Public Eye publishes analysis of it.

For some individual crops it is tighter still, and a handful of firms hold most of the relevant patents.

The licensing will look familiar if you have ever read a software agreement.

A patented seed is not simply sold. It is licensed, and the terms can stop the oldest practice in farming: keeping part of this year’s harvest to plant next year.

So the farmer buys the use of the seed for 1 season. The right to reproduce it, which is the whole point of a seed, stops with the supplier.

That is a permanent purchase turned into a recurring one. It happened to farming before it happened to software.

There is a second familiar detail.

The seed is often built to work with a particular pesticide, and the same company sells both. Buying one makes buying the other easier than switching.

Agricultural economists have tracked the results: fewer varieties about, higher input costs, and less ability to change supplier.

Supermarkets, and the rules built for them

The other half of food shows the same problem from the buying side. This is the example worth copying.

A small number of retailers handle most grocery sales in the United Kingdom.

So a supplier faces very few possible customers. Losing one can finish the business. For the retailer, replacing a supplier is a morning’s work.

That imbalance got documented well enough that Parliament acted. The Groceries Code Adjudicator was set up in 2013. It oversees whether large retailers follow the Groceries Supply Code of Practice.

Look at what that code restricts, then think about your last software renewal.

  • Changing agreed terms after the fact, without agreement.
  • Charging a supplier for the right to carry on doing business.
  • Moving costs and risks onto the supplier without compensation.
  • Using removal from sale as leverage in a dispute.

The European Union built its own version. Directive 2019/633 on unfair trading practices covers the agricultural and food supply chain.

It bans late payment, cancelling orders at short notice, changing terms one-sidedly, and threatening commercial retaliation. Every member state has an authority to enforce it.

Neither regime is perfect. The Adjudicator’s powers are limited, it does not cover pricing itself, and it has used its strongest powers sparingly.

But the principle inside them is the one missing from technology.

Where bargaining power is very unequal, freedom of contract does not really exist. So certain practices get banned outright, rather than left to negotiation.

Nobody has suggested a software supplier should be stopped from changing agreed terms after the fact.

Nobody treats a charge for taking your own data out as shifting a cost onto the weaker party.

In food, both would be spotted straight away. In software we call it a licensing model.

Why technology has nowt like it

Four reasons, and they are worth understanding because they point at what would have to change.

Speed. Grocery concentration took about 100 years. Cloud concentration took about 15. By the time you could see it clearly, it had happened.

Free services. Competition law in most countries grew up round prices paid by consumers. A service priced at zero looks harmless under that test, even when the customer is not the user.

The claim of novelty. The industry argued, successfully, that its economics were different, and that being hard to leave was just a property of complex systems rather than a design choice.

Organisation. Farmers have unions, cooperatives and a long habit of bargaining together. They have used it to win specific legal protections.

Technology buyers have user groups and a conference. When VMware customers were hit with large increases, the real pushback came from a trade body of cloud providers, and it went through competition law. Part 2 covered how long that takes.

What it tells you

Two conclusions, and both are worth having.

The problem is structural, not national.

Bayer is German. The biggest supermarkets are British, French and German. The buyout model in part 3 is used keenly across Europe.

Replace every American supplier in this series tomorrow with 4 European ones and the behaviour comes back, because it follows from the structure.

As such, the advice here is about being able to change supplier, rather than about picking a country.

A working answer already exists.

We do not need to invent a new theory of platform regulation.

There is a model where certain practices are banned because bargaining power is unequal, an adjudicator hears complaints, and the weaker party does not carry the whole burden of proof.

It was built for cabbages. It would work on cloud contracts.

Part 8 looks at what is being built in Europe now, and how to check where you stand.

First published: 2026-08-25. Last updated: 2026-08-25.