Review of the Turing Tumble puzzle as a potential classroom activity
Created by Richard Pawson
last edited Feb 09 2020 by Richard Pawson
I recently took delivery of a Turing Tumble.
I was intruiged by this idea because I have always been a fan of problem-solving-by-construction, both in physical form (e.g. Meccano, Lego) and as computer software. Bill Budge’s Pinball Construction Set, released in 1983 for the Apple II, had a huge impact on my thinking about software design, and The Incredible Machine (arguably, the PCS’s spiritual successor ten years later) had a major impact on my career. (If that last statement intrigues you, you might enjoy this presentation).
Turing Tumble appeared to be a physical incarnation of the same idea. It is certainly enjoyable - leaving aside my grumpy-old-man dislike of the Manga-style instruction manual. The set puzzles are enjoyable, and range from straightforward to extremely challenging. The website promotes the use of Turing Tumble in the classroom for teaching the principles of computing. I’m not really a fan of teaching ‘computational thinking’ in the abstract, but if I had to do it, this would seem attractive.
But I do have some quite strong criticisms of it, in two areas.
The first is the physical implementation, which is, IMO, far more flimsy than it should be. Within five minutes of starting, I had accidentally knocked the stand-up board with my hand, resulting in a dozen of the small (7mm) hard balls bouncing across my (also hard) floor and into various crevices. Sometimes, in normal use, balls would bounce out of the frame, or by-pass elements in a non-deteministic fashion. Occasionally some of the mechanical components would be a bit ‘sticky’ on their pivots.
These faults, could well be rectified in time. Perhaps, as with many Microsoft products, the best advice is to ‘wait for version 3’. But as it stands, I could not imagine this surviving long in the classroom.
This situation is made worse by the fact that there is already more than one excellent software emulations of the Turing Tumble puzzle, e.g. here. These are more extensible and free to use!
I like the idea of the physical implementation - to get kids off their screens - but the current implementation isn’t good enough.
The promoters claim that the puzzle is ‘Turing complete’. Certainly you can construct some very interesting devices to solve computational problems with it. But is it really Turing complete? I found myself rather sceptical about this claim.
I don’t think this is being pedantic. If you are going to sell a product, aimed in part at computing eduction, and you claim it is ‘Turing complete’ then that claim should be demonstrably true.
There is quite a substantial discussion of this question here, to which I have joined my voice. I remain on the sceptial side, though am certainly open to being shown to be wrong on that front. The discussion gets quite technical, but there are some intriguing ideas, and very sophisticated designs put forward.
The most disappointing thing about that discussion is that it shows, from the inventor’s own participation in it, that he had made that claim without giving it much serious thought, let along having any proof or rigorous argument in support.
It irks me because I think that the treatment of Turing Machines in the A-level exam spec is poor, and perpetuates some serious misunderstandings. This puzzle certainly isn’t going to help.
In conclusion, while I’m glad to have had the opportunity to try it, my Turing Tumble will shortly be going on eBay!