ENIAC ....

On 2011 Jan 25, at 3:33 PM, Evan Koblentz wrote: Yes, one does has to be careful about Wikipedia. To be clear, I wasn't using wikipedia as a proof or evidence source. I had never seen ENIAC characterised as Turing-complete, I thought there was some limitation to it's programmibility and addressability that would inhibit it, there are certain requirements for a machine to be Turing-complete. Although as I alluded with the Minsky ref, I'm not entirely surprised that it _might_ be possible. I will nonetheless wait for the proof or explanation to give it credence, as well as see how tortured the characterisation is (what it takes to accomplish it). I do have to wonder if whoever made the characterisation is confusing Turing-complete with a notion of programmibility or 'general-purpose' (another ambiguous phrase). The ENIAC page links to the "Turing completeness" page. There, Conway's "The Game of Life" is stated to be Turing-complete. The point being, it is interesting to look at these things from a strict or theoretical point of view, and there is a wide range of things that can be characterised as Turing-complete. So not to take away from the Turing-complete matter (I introduced it), but let me put it another way: there is a very large distinction between the architecture and practical capabilities of the ENIAC (or an ENIAC-like machine) and the stored-program machines that came after. Or let me express it a little more practically: I can reasonably imagine writing an Algol compiler or a word processor to run on the EDVAC, even without knowing the instruction set of the EDVAC. Even the lack of character handling could be worked around without much difficulty. Doing so on the ENIAC is another matter, and I'm allowing for having a large number of accumulators (data memory), as one would in a Turing-completeness assessment. If it's Turing-complete, yes, one can do it, but how tortured is it? -- There are some curious things going on in the wikipedia "Turing completeness" article. Looking at the revision history, the article starts out with no ref to ENIAC, then this is added: "The first machine known to be Turing-complete was ENIAC" Later, to account for the paper on the earlier Z3 it becomes: "Previously, the first machine known to be Turing-complete was ENIAC" Somebody changes (reverts) that to: "Globally however, the first machine known to be Turing-complete continues to be ENIAC (1946)" Whatever that means. In Feb 2010 (and currently) it becomes: "but the first machine explicitly designed to be Turing complete and widely appreciated as being universal was the 1946 ENIAC. This machine was able to solve a wide range of effective problems in the 1940s, many related to atomic bomb design." That first bit is a rather strong statement of intent. Strikes me as a little bit of revisionist history. I'd be rather surprised if Mauchly and Eckert knew what Turing-complete meant when they designed the ENIAC. The second sentence is a nice promotional statement about the ENIAC, its presence in the Turing-completeness article is questionable.