Rickover's Memo as a Dunning-Kruger Curve

By Nick Touran, Ph.D., P.E., 2025-06-26 , Reading time: 2 minutes

Here’s the Dunning-Kruger effect as envisioned by Rickover. Many variations of this curve exist and have been traversed.

Dunning-Kruger curve for nuclear reactor development

The UNGGs, Magnox’s, and RBMKs all made it to commercial fleets, but were then replaced with higher-performance reactors (mostly PWRs).

LWRs made it commercial, but made major modifications and cost adjustments after unexpected fleet incidents happened, like the Salem ATWS, Browns Ferry Fire, TMI, and Fukushima. Even before that, many retrofits added to the initial delivered cost (e.g. San Onofre 1 where they put another containment around their containment).

SFRs reached commercial demo a few times and petered out, first with the dismal Fermi-1 but then other times with Dounreay, SNR-300, CRBRP, and SuperPhenix. The BN-800 lives on in Russia, China has CFR-600s, and India almost has the PFBR. In the US, several companies are working to get to demos online.

HTGRs got to first commercial demo a few times and petered out with THTR-300 and Ft. St. Vrain due to poor performance. But nowadays China had success with demo HTR-PM and is building the 2 6-packs of HTR-PM600s! Many other HTGR orgs are rushing to get their first demos online.

For Hallam and Piqua, the first commercial demos weren’t good enough to justify continuation, and the programs were closed down. Maybe they will be revived? (I love these reactors!)

For MSRE, the first demo wasn’t good enough to convince the funding parties to continue the program. (As a rule, the reactor developer and their allies always believe the reactor program should have been continued). Decades later, China built TMSR-LF1 to try again (and many others are trying to build).

Perhaps CANDUs have been the least impacted by negative learning? They’ve evolved a bit and continue delivering good performance.

Dozens of other reactor programs either never made it to first demo or had their program closed down after the initial demos failed to impress (e.g. HRE-2, UHTREX, LAMPRE, ARE, superheat BWRs, etc.) What does your curve look like and where are you on it?

See Also


Reader comments