Konrad Zuse/Talk

< Konrad Zuse

HomePage | Recent changes | View source | Discuss this page | Page history | Log in |

Printable version | Disclaimers | Privacy policy

There has been considerable debate over whether the Z-series machines really constituted computers on the history of computing/talk page and the associated timeline page.

The consensus is leaning towards the view that, while they represented something close to a computer and had many interesting features, they didn't have looping constructs which is a prerequisite for a real computer.

I think there is no such thing as a first computer, any definition would be arbitrary. All calculating stuff from this time is groundbreaking and should be named 'among the first' or something similar. If enough details are given, anyone can make up his own mind. --Yooden

I disagree. To my mind, there is a fairly clear definition possible - a computer should have equivalent capabilities to a universal Turing machine (modulo storage capacity). From what I've read so far, none of Zuse's machines had this capability, whereas ENIAC (for example) almost certainly did. -- Robert Merkel

OK, you may be right, I'm lacking theoretical knowledge here. In this case, I suggest to write Definition of Computer or something similar and point to it whenever appropiate.

I just checked Computer and the definition there. It uses a Von Neumann model which is mentioned in Konrad Zuse. Could you clear this up? --Yooden

The article Computer needs quite a bit of editing (I hadn't seen that one before, I'll clean it up at some stage). The "Von Neumann model" is a model of the internal organisation of machines that are computers. All single-processor machines currently in use fit this model (multi-processor machines of various types vary from it lesser or greater amounts IIRC). Zeus's machine does not implement a full Von Neumann model, as the program is stored in a logically separate form of memory (paper tape, which was read-only as far as the computer was concerned) to the data (stored in an arrangement of shifting metal bars).

Regardless, whether Zuse's machines were Von Neumann architectures is irrelevant to the question of whether they are computers or not. A Von Neumann computer describes a *design* of a computer, not their capabilities - and capabilities are really what I'm discriminating on here.

However, an article by Zuse's son http://www.epemag.com/zuse/ points to a 1998 scientific paper in an IEEE journal supposedly proving that the Z3 did in fact have the capabilities of a Turing machine. I find this difficult to believe, as the Z3 lacked a type of instruction called a "conditional branch" - in essence, an instruction saying "if (proposition X) is true, go to point (Y) in the program, otherwise keep going to the next instruction. It is difficult to imagine how you could get equivalent capabilities to a Turing machine in a machine without a conditional branch instruction or something similar, but IEEE generally doesn't publish papers with wrong proofs in them. Anyway, I'll go to my old university library and dig up the paper, and if it checks out Mr Zuse's Z3 should be regarded as a computer (though perhaps only in a theoretical sense - it may be that the machine may have been thoroughly impractical to use as a full-blown computer) in my opinion. Robert Merkel

I have seen the paper; with a trick that Zuse never anticipated and that is thouroughly impractical, Z3 can be made Turing complete. --AxelBoldt

capabilities are really what I'm discriminating on here.

Pray do discriminate, you seem to know a lot more than I do; but please do it explicit, like 'Zuse was almost there, but the Z-3 lacked somesuch, so Colossus Mark I is the first computer'. (The fact that Zuse is German does only affect my knowledge about him; I don't care who's first.) --Yooden

Colossus *wasn't* a general purpose computer - there's agreement on that. It was a fascinating electronic device, but it was a special-purpose machine that could do one thing and one thing only. Mark I doesn't meet the definition I've given either. However, the Z3 does - but with the catch that it wasn't known that it could do so until 60 years after it was built and by which time there were many millions of computers meeting the definition in existence.

Zuse built a machine that could have been used as a computer (meeting the Turing-complete definition) before anyone else did, but it's totally impractical to use it as such, Zuse never figured out how it could be done, and indeed nobody knew it was the case until three years after he died and by which time computers millions of times more powerful were available for negligible cost on a single chip. That leaves us with:

  • The analytical engine would have been a computer, if ever built.
  • Zuse's Z3 was a computer, but nobody knew it was until nearly 60 years later, and it wouldn't have really been practical as such.
  • ENIAC was a computer.
  • EDVAC was the first practical stored-program computer of the architecture we know today.

That seems to be the state of play to me.

So, after all that, Yooden's quite right - it's very difficult to anoint anybody with the title of "inventor of the first computer" or a particular machine as "the first computer". However, we should try and edit this information here in to the relevant articles. RM

One caveat: I think EDVAC was the first design of a stored program computer, but the first functional stored program computer was the Manchester "baby" of 48. --AxelBoldt

However, we should try and edit this information here in to the relevant articles.

Before I get my hands dirty, I would ask you to update Computer, so that I can rely on that. --Yooden

Shall do. To clarify what I mean by Turing-equivalent my definition, you might want to look at Turing machine, particularly the idea of a Universal Turing Machine. When I say a machine is "Turing-equivalent" or such, it means that the particular computer is capable of acting as a universal Turing machine, except that its storage capacity is limited. There is an important precept of computer science (it's not a theorem because it can't be proved, only disproved, but everybody believes it to be more than likely true) that says that *any* other computer-like system you can construct can be emulated on a Universal Turing machine. So, if you have a computer capable of being a Universal Turing Machine, you have a computer that can do anything any other computer (in the broadest possible sense) can (given enough time and assuming we've got enough storage space). This is why Turing-equivalence (usually called Turing-completeness) is a natural criterion for computer scientists to choose for determining whether a machine is a "computer" or not.

That certainly doesn't sound arbitrary, and Turing machine may serve as a reference.

Please add 'Turing-equivalence' and 'Turing-completeness' at the right places in Turing machine. --Yooden

The last change is referring to HWR's comment on History of computing/Talk. Details anyone? --Yooden

If you mean the simulation of a Turing machine by the Z3: it's at http://www.inf.fu-berlin.de/~widiger/ICHC/papers/universal/universal.html --AxelBoldt

The Z-3 was built as a prototype to test the feasibility of a machine capable of solving the large systems of linear equations required for analysis of resonance in airplane wing designs in order to avoid the problem of aerodynamic "flutter". The limited memory of the Z-3 was not capable of handling the large problems where manual methods were impractical, so it was never used for any practical purpose, only small tests. This was of course the intention. The success of the Z-3 led to the authorization of the full-scale machine, the Z-4, but this machine was not completed before the end of the war, due largely to the impact of Allied bombing raids on Berlin after 1944. Zuse also built a small relay device called the S-1 which executed a fixed sequence of calculations. It was used by the Henschel Company in the production of the HS-293 flying bomb, which was used in large numbers toward the end of the war, against Allied shipping in the Mediterranean and in the German retreat from Poland.

Source: Paul E. Ceruzi, Reckoners (1983) - HWR

Sounds good; put it in where I only hinted at the guided missile. --Yooden

I think you oughta add some things from his autobiography: 'the computer, my life' or whatever its called.

1. he did a hell of alot of his work in a converted room in his parents house

2. he and his buddies were a bit dumbfounded as to what to do about the racism of the government, they seem to have had no idea about what to do about it.

3. when he was trying to hire helpers for his company, the only person he coudl get was a blind person, because under the nazi regime all 'strong' people were devoted to military service, while the 'defectives' such as the blind were considered unusable. However the blind person easily understood the concept of binary arithmetic and helpd him assemble his machines.

4. He tried to get the military interested in his machine to help airplanes, but the guy he talked to basically said "why should we want to improve our planes? they are already perfect"

5. he had a hell of a hard time selling his machine to businesses after the war. they didnt see the point of using his machine.

To the person who added the above, why don't you add this stuff into the article where it fits? Much of it seems

relevant, though you might want to change the tone a little --Robert Merkel