George Dyson's history of the Colossus computer (Nature 482, 459–460; 2012) is somewhat misleading. The development of Colossus owed little to the Bletchley Park 'bombe' devices.

After Bill Tutte developed a statistical strategy to tackle the German Fish cypher, Max Newman proposed a machine to implement it, using two paper tapes. With help from the Telecommunications Research Establishment (later the Royal Radar Establishment) and from Frank Morrell's group at the Post Office Research Station at Dollis Hill, the Robinson machines were built. These were slow but validated the technique. Thomas Flowers' group recommended the use of many more vacuum tubes to improve performance, but they were wrongly considered too unreliable by Bletchley Park.

So Flowers built the first Colossus at Dollis Hill instead of Bletchley Park although, in time, Newman spoke to him about his requirements.

One crucial technique developed by Turing and used by Tutte was that of 'delta-ing', or using the differences between characters rather than the characters themselves. If Dyson had pointed out that Tutte used this method to detect the statistical non-randomness of plain text, it would also have helped in understanding how different Fish was from Enigma.