Forums  > Off-Topic  > The MONIAC; a macroeconomic computer using fluidic logic  
     
Page 1 of 1
Display using:  

jslade


Total Posts: 1206
Joined: Feb 2007
 
Posted: 2010-07-09 08:55
I've always been a fan of non-standard computer architectures. At one point, I had thought about trying to make a career of attempting to make a useful one. Anyhow, I came across the MONIAC again recently, and thought others might be interested in this bit of quanty history.

http://en.wikipedia.org/wiki/MONIAC_Computer

some cool photos

quick old school video

more detailed video

Pretty inspiring what you can do with limited models.

"Learning, n. The kind of ignorance distinguishing the studious."

urnash


Total Posts: 566
Joined: Sep 2006
 
Posted: 2010-07-09 09:33

Cool. Thanks for sharing!

Although you can argue whether this is a computer. I'd say it's a model. Or a simulation environment, or (can't think of a better word right now, sorry).

IMHO a computer should be general purpose, and AFAIK this isn't. I know that with this definition many of the old mechanical computers would not be classified as a computer. But I'd say that e.g. the Z1 was a computer, but the MONIAC wasn't. Note that this is based on 5 whole minutes spent looking at MONIAC websites...


understanding shit will always be worth something -- filthy

rod


Total Posts: 398
Joined: Nov 2006
 
Posted: 2010-07-09 11:02
A computer is a machine that computes. Before the age of general-purpose digital computers, people used to solve differential equations using resistors, capacitors and op-amps. They were specific-purpose analog computers, but they were computers nonetheless.

Some more links on the MONIAC / Phillips Machine:

Can control theory save the economy from going down the tubes? (2009)

Phillips machine: like water for money (2009)

Phillips machine: the computer that once explained the British economy (2008)

urnash


Total Posts: 566
Joined: Sep 2006
 
Posted: 2010-07-09 12:57

OK, so we can say that there's a difference between a General Purpose Computer and a Computer. However, according to Rod's definition, is a very simple pocket calculator also a computer? And what about Pascal's calculator?. I'd say that there's a difference between computing and calculating. But you could call that semantics..

BTW, rod's second link poits to an article that contains the schamtics of the MONIAC. That helped me to understand how it worked much better than all the pictures...


understanding shit will always be worth something -- filthy

rod


Total Posts: 398
Joined: Nov 2006
 
Posted: 2010-07-09 13:13
urnash,

Aren't computer and calculator synonyms? It all depends on what definitions one uses, of course. I have no problem with calling computers to general-purpose machines only... just as long as one keeps in mind that there are non-electronic and specific-purpose computers as well. Hell, I have seen Turing Machines built with Lego. I wouldn't use them to solve differential equations, though.

urnash


Total Posts: 566
Joined: Sep 2006
 
Posted: 2010-07-09 13:15

Let's agree that it's a grey area. I wonder what somebody who's a native English speaker makes of the distinction between the two.

One of the Moniac's was sold to the bank of Guatemala. An artist made a "project" out of it, see http://www.michaelstevenson.info/projects/answers_to_some_questions_about_bananas/. There's even a picture there of Professor Meade teaching global economics at the LSE with his two nation model, ie. con-joined twin Moniac machines! Source is your nytimes blog link.


understanding shit will always be worth something -- filthy

jslade


Total Posts: 1206
Joined: Feb 2007
 
Posted: 2010-07-09 19:37
urnash: "Although you can argue whether this is a computer. I'd say it's a model. "

I'm not sure how the semantics of all this boil out, but as rod pointed out, there did used to be an active analog computing field, and MONIAC certainly qualifies as one of these. They were useful enough people sold them until fairly recently. They were also used as "math coprocessors" in early digital computers, which were often too slow to efficiently calculate things like integration of nonlinear differential equations. Some Israeli lady claimed to come up with an analog computing architecture which was super-Turing in the late 90s, but I suspect it was a repeat of the old "there is lots of bits in real numbers" error: nothing ever came of it anyway. You could probably think of quantum computing as a sort of weird digital/analog computer, where the bits are complex numbers.

Back in the 50s and 60s, computer science was a real science. There were lots of alternatives to Von Neumann machines. MONIAC is one example of a very practical one. Lots of the WW-2 fire control computers more or less resembled this idea, which is probably how it got built. If you have an extra hour or so, here is a series from the US Navy on programming electromechanical analog fire control computers. My favorite weird architecture for computers: stochastic computing. A sort of hardware implementation of Monte Carlo. They built them that way, in part, to fit into the noisy environment of early missiles: the noise ended up being part of the calculation! Anyway, virtually all of this is forgotten technology now.

"Learning, n. The kind of ignorance distinguishing the studious."

cowpoke


Total Posts: 92
Joined: Jul 2009
 
Posted: 2010-07-09 20:15
That's very cool. It looks like it can't simulate government debt. Very cool thing, though.

urnash


Total Posts: 566
Joined: Sep 2006
 
Posted: 2010-07-12 08:36
jslade: thanks for the links!

understanding shit will always be worth something -- filthy

HankScorpio


Total Posts: 479
Joined: Mar 2007
 
Posted: 2010-07-12 16:18
from jslade's link
http://upload.wikimedia.org/wikipedia/commons/7/7e/AKAT-1.JPG
how cool is that. Analog computer by day, party machine (mixing desk) by night! Smiley

[Edit:] changed image to link, as image was of gigantor size.

urnash


Total Posts: 566
Joined: Sep 2006
 
Posted: 2010-07-13 13:24

Sorry for boring all of you, but I try one more time:

jslade: I'm not sure how the semantics of all this boil out, but as rod pointed out, there did used to be an active analog computing field, and MONIAC certainly qualifies as one of these. 
 
I disagree. There were certainly analog computers. Even I count them as a computer (e.g. the GP-6 that you point to). But I don't see how MONIAC is one of them. As far as I can see you could not program MONIAC. You simply fed it inputs, and watched the outputs, but you could not change the program. And being able to change the "program" that the computer runs is what IMHO makes a computer a computer. And there is a distinction between changing the program and changing the input parameters (even though there is a grey are between the two).

BTW jslade, your 'Israeli lady' link does not work. Do you have a different link?


understanding shit will always be worth something -- filthy

Dizzy


Total Posts: 251
Joined: May 2006
 
Posted: 2010-07-13 13:32
Urnash,

I think jslade referred to this story:

http://www.cs.umass.edu/~binds/docs/EETimes.pdf

Pray, Mr. Babbage, if you put into the machine wrong figures, will the right answers come out?

rod


Total Posts: 398
Joined: Nov 2006
 
Posted: 2010-07-13 20:05
The Israeli lady that jslade mentioned is Hava Siegelmann. A couple of papers she wrote:

Analog Computation with Dynamical Systems [pdf]


A Theory of Complexity for Continuous-Time Systems
[pdf]

But the idea of studying the complexity of analog computation is not new. For instance, Ken Steiglitz wrote a paper on that in the mid 1980s:

The complexity of analog computation [pdf]

In the late 1980s / early 1990s, Roger Brockett wrote a bunch of interesting papers on analog computation, too. For instance, he designed systems of nonlinear ODEs that could sort lists, diagonalize matrices, count pulses, etc. It's fascinating stuff.

jslade


Total Posts: 1206
Joined: Feb 2007
 
Posted: 2010-07-13 23:03
urnash: I'm pretty sure you can move pipes around, and I'm certain you can alter flow rates and turn things on and off and so on. One of the links discussed hooking a couple of differently configured MONIACs to model two economies interactions. Not any different than the patch board on the GP-6, IMO.
Rod and Dizz pointed to the right "Israeli lady" and her super-Turing research; sorry for the malformed link.
Also, thanks for that Steiglitz paper, Rod. I was very interested in spin glass computing (though I didn't find that paper in my old bibtex); spin glasses and certain dynamical systems map very easily onto NP-hard problems. It's all very interesting stuff, and makes me wish I had worked harder on it in those lazy academic days.

"Learning, n. The kind of ignorance distinguishing the studious."

rod


Total Posts: 398
Joined: Nov 2006
 
Posted: 2010-07-13 23:21
Programming an analog electronic computer would be to calibrate it, i.e., tweak the potentiometers until one obtained the desired new dynamical behavior (i.e., one would calibrate in order to change the natural modes / poles / eigenvalues of the dynamical system). Some more stuff on analog computers:

MIT Differential Analyzer

A great disappearing act: the electronic analogue computer [pdf]

And some of Brockett's papers that I mentioned before:

Dynamical Systems that sort lists, diagonalize matrices and solve linear programming problems (1988) [pdf]

Smooth dynamical systems which realize arithmetical and logical operations (1989)

Analog and Digital Computing (1992)

Pulses, periods, and cohomological terms in functional expansions (1995)

Dynamical Systems and Computational Mechanisms (2002) [pdf]

urnash


Total Posts: 566
Joined: Sep 2006
 
Posted: 2010-07-14 09:04

jslade: and I'm certain you can alter flow rates and turn things on and off and so on. One of the links discussed hooking a couple of differently configured MONIACs to model two economies interactions.

Ahh. I've missed that. Didn't know that that was possible. Now I agree that the MONIAC was a computer.

rod & dizzy: thanks for the links!


understanding shit will always be worth something -- filthy

nikol


Total Posts: 977
Joined: Jun 2005
 
Posted: 2020-01-07 21:23
Just rediscovered the subject.

In particular, I like the term Hydraulic Macroeconomics
https://en.wikipedia.org/wiki/Hydraulic_macroeconomics

This thread is not Off-topic. It deserves to be part of Pricing and Modelling.


Thinking aside: Assuming that fluids cannot be compressed/inflated, can this model also cover inflationary features invented later by many CB's for their benefit?
Previous Thread :: Next Thread 
Page 1 of 1