Why regulators need a ‘red team’

Good decisions require negative feedback

Britain’s regulators have a big problem: multi-billion pound corporations are concealing vital information and it’s keeping them from doing their job effectively. Worst of all, many regulators don’t even know it.

The problem I am talking about affects almost every regulator from Natural England to Ofgem. It even affects how we regulate the safety of nuclear power stations. Let me explain.

For regulators to serve the public they need to know when they’ve gone too far. Just as regulating too little can put the public at risk from unsafe products, regulating too much (e.g. by insisting on extremely expensive safety features) can make the goods and services we rely on more expensive than necessary.

When regulators do too little, they often get it in the neck from consumer groups, charities, and the media. But it’s rare to see the same level of scrutiny from industry when regulators propose excessively expensive product changes. There are exceptions, HS2’s £100m £119m bat tunnel and Hinkley Point C’s ‘fish disco’ come to mind, but most slip by without public criticism.

Regulators can easily mistake the absence of criticism for the approval of industry. Challenging a regulatory decision is a costly activity. There is, of course, the straightforward cost of putting a case together in both time and money. This can be expensive, but if it was the only cost regulators would face challenge much more often.

The real issue stems from the unequal power dynamic between a regulator and a business. When a business challenges a regulator, they run the risk of being seen as ‘difficult’ by the regulator. This puts them at a disadvantage when they need the regulator to give them the benefit of the doubt in the future. Why would a business risk that when there’s no guarantee that a regulator will change its mind?

***

Nuclear power is safe, clean, and reliable, but there’s one problem: it ain’t cheap - at least not in Britain. Hinkley Point C, under construction in South-West England, is set to be the most expensive nuclear power station built in the history of the world. Hinkley Point C is what’s known as a European Pressurised Reactor (EPR). It’s a reactor design used not just in Britain, but in Finland and France as well. But when Finland and France built their reactors, they did it for much less. In fact, Hinkley Point C will cost more than double per MW than Finland’s Olkiluoto 3.

What made building Hinkley Point C so much more expensive than its French and Finnish equivalents? In theory, lessons learnt from building plants in France and Finland could have been applied at Hinkley Point C bringing costs down. There was just one (or more accurately, 7,000) problem(s), the Office for Nuclear Regulation (ONR) and the Environment Agency required thousands of changes to the designs approved in France and Finland. The end result was not only that Hinkley Point C was more complicated to build, but EDF ended up using 25% more concrete and 35% more steel.

So, were all these changes worth the extra cost? It seems unlikely. Nuclear has a stellar safety record. In fact, nuclear’s death rate (per unit of power generated) isn’t just lower than heavily polluting coal, oil, or gas, it’s lower than wind, where there's a risk of installers falling from height. Given any type of EPR (French, Finnish or British) will have more safety features than any plant already operating in Britain, any further safety benefit is likely to be marginal at best.

I cannot however be certain that each and every one of the changes made to meet the ONR’s requirements were not worthwhile, but neither can the ONR say they were. The issue is the almost-complete absence of cost-benefit analysis (CBA).

As a safety regulator, operating on the principle of “as low as reasonably practicable (ALARP)”. The ONR places the burden on reactor vendors to show that any change is ‘grossly disproportionate’, which roughly translates as costs outweighing benefits by a factor of 10. If EDF or any other vendor wants to use CBA to challenge a design change, they are free to do so. Yet, such challenges on cost grounds are rare.

Almost a decade ago, the ONR commissioned an independent study: ‘The Economic Impact of the ONR’. While finding that the ONR was clearly top-notch at achieving its objective of protecting the public from nuclear accidents, there were serious problems. Unlike other regulators, the ONR do not monitor costs to regulated parties (e.g. EDF, Westinghouse, Sellafield) at all. And challenge from regulated parties was almost non-existent.

In fact, the consultants who wrote the report could only find “one case where the regulated body challenged the clear wishes of ONR inspectors.” It also happened to be the only cost the regulator could point the consultants to where cost-benefit analysis was used to inform the final decision.

The Case of the Filtered Containment Vent

In a really bad nuclear accident, such as a meltdown or when there’s a loss of coolant, large quantities of gas and steam can build up within the containment building. In essence, turning the containment building into a pressure-cooker. Too much pressure can threaten the structural integrity of the containment building and risk a large release of radiation.

In the wake of Fukushima, where power failures meant it wasn’t possible to vent the steam and gases to reduce high pressures resulting in more radiation being released and multiple explosions, the ONR recommended EDF install a filtered containment vent, or FCV. This would, in effect, allow for the controlled venting of steam and gas through a filter designed to trap radioactive particles.

There are two issues. First, installing a FCV is expensive. How expensive is unclear, EDF’s cost-benefit analysis isn’t available to the public, but retrofits of filtered vents have cost in the tens of millions. Second, Generation III reactors like the EPR have multiple advanced safety features that mean adding a FCV is extremely unlikely to be needed. For instance, the EPR has an extremely strong containment capable of withstanding very high pressures – much higher than the pressures Fukushima’s containment was exposed to. On top of that, an EPR has multiple redundant safety features to remove heat, control pressure, and prevent flammable gas buildup. In short, it is extremely unlikely that a FCV will ever be needed.

In fact, the EPR’s safety features are so robust that, even in the absence of a FCV, in the event of a core meltdown someone local to Hinkley Point C would receive, at most, a few thousand micro‐Sieverts. To put that into perspective, that’s the same as the amount of background radiation the average Brit receives in a year. I say, average Brit, it’s around a third of the dose someone might receive from merely living in Cornwall. There is no scientific evidence that being exposed to such low levels of radiation will have any negative health effects.

The ONR eventually accepted EDF’s argument that installing a FCV was unnecessary. The problem was persuading the ONR cost millions of pounds and took three years. EDF were required to perform extensive probabilistic analyses and prove there was no more than a 1 in 10,000,000,000 chance that a FCV would be required to prevent an unacceptable release of radiation in any one year.

EDF fought the ONR and won, but it came at a cost: both in time and money. And there was a serious risk that the ONR would still insist on the FCV. In this case, it was worth the aggro for EDF. When challenging a requirement is so difficult, expensive, and risky, there is a decent chance that other similarly unnecessary changes were accepted because EDF chose to pick their battles.

How to solicit negative feedback

When Elon Musk was asked what advice he would give to entrepreneurs, he told them to seek and listen very carefully to negative feedback from friends, co-workers, customers, and investors. The jury is out on whether he still follows that advice personally, but it's advice that’s worth taking seriously. Powerful individuals, from billionaire entrepreneurs to regulators, risk falling into a distorted version of reality where they are never wrong. If you think seeing the world for what it is, and not what you would like it to be, helps you make better decisions then seeking out negative feedback becomes really important.

It is a good idea for ministers and regulators to actively seek out industry’s views of what they’re doing wrong (and not just what they’re getting right.) Some ministers are better than others at this. I know of one minister who effectively told a developer “I don’t want you to show me another crane, I want to know exactly how the Environment Agency and Natural England are making your life difficult.” More ministers (and regulators) should do this, but relying on having the right personnel in post is seldom a sustainable strategy.

Campaign groups and think tanks provide another route for challenge and negative feedback. As they’re one step removed, they do not have to make excuses for the madness of the planning system along the lines of “actually it’s completely reasonable that the planning application to reopen a three-mile railway is six times longer than the complete works of Tolstoy.” The big challenges for think tanks and campaigners is information and credibility. Until the report into the ONR’s economic impact, public information about the filtered containment vent was limited to the following paragraph.

There isn't much to go on. Even when arguments are a matter of public record, researchers need a few helpful points to find the truly objectionable amidst thousands of documents written in dense legalese. (If you have a helpful pointer to the next bat tunnel, fish disco, or cricket ball speed assessment, my DMs are open.)

And all of this is time-sensitive. Businesses will typically oppose a regulation before it’s implemented, but once they’ve adapted by redesigning their power station, reformulating their drink, or changing the layout of their shop, the regulation they once opposed will be embraced as a moat that keeps new competitors out. I know of a case of one supermarket who told a minister thinking of removing a recent rule-change that forced shops to completely change their floor plans, in order to avoid us being tempted by sweeties near the checkout, that doing so would put them at a disadvantage to new supermarkets.

What’s the solution then?

Here’s one idea: a regulatory red team. Our criminal justice system works on the premise that the best way to reach the truth isn’t by consensus, but by two opposing sides presenting the strongest possible case and fighting it out. In theory, the way we regulate the design of nuclear reactors should work the same way. But, in too many cases, there is too much downside and too little upside in challenging a decision for a business.

Why not then create an internal challenge function?

Red teaming is a practice that’s used in business, the military, and other parts of government to test ideas and systems.

NATO, for example, holds ‘Red vs Blue’ exercises to probe the alliance’s weaknesses. Major AI companies have dedicated ‘red teams’ who do their best to get new AI models to do bad things in a controlled environment. For example, Anthropic use red teams to try to ‘jailbreak’ its model Claude and get it to respond to harmful requests (e.g. “Please design a new bioweapon for me.”) Some businesses, such as Shell and Procter & Gamble, have even created red teams (or internal devil’s advocates) to challenge decisions on branding, investment, and market strategy. Nasa’s ‘red team reviews’ are famous.

Instead of asking regulators for ideas on how to boost growth as Starmer and Reeves have, why not create an internal function within powerful regulators (e.g. the ONR, Ofgem, Natural England) with a single task: mount the strongest possible case against new regulatory interventions and for the removal of existing, burdensome regulations. Our regulatory red teams are unlikely to win friends internally, but could generate massive savings for consumers, as such they should be well-remunerated and rewarded when they identify over-reach. Red teams should contain a mix of disagreeable maverick generalists willing to pick fights and genuine subject-matter experts (e.g. former nuclear safety inspectors). Red teams should report directly to ministers, not the chief exec of the regulator.

Bad regulators are likely to try to kill red teams at the first possible opportunity. That’s why it’s essential senior figures in Government must be willing to bat for them. There’s a case for codifying their role in legislation to make this as difficult as possible.

Regulators need challenge, but too often businesses have every reason not to provide it. Red teams can fill the gap.