Hacker News new | past | comments | ask | show | jobs | submit login
Mechanical Turing Machine in Wood (youtube.com)
68 points by ColinWright 9 months ago | hide | past | favorite | 22 comments



This is fascinating. It reminds me of an art display I saw years ago that looked similar. However I was disappointed to learn that in that case the mechanical stuff was only for show and the algorithm was driven by an Arduino hidden behind the scenes. I'm so happy someone actually made the real thing!

Don't miss the document linked in the video description. It gives some details on how the machine works.

The alphabet on the tape has 3 symbols: "b", "0", "1" - they are encoded as different positions of pegs in the physical tape. Three non-functional tape blocks mounted on the machine demonstrate the encoding of the alphabet.

There are 3 machine states "a", "b", "c". They are encoded as the major position of the plate on the left of the machine that moves up and down. The plate also has a minor position that depends on the current tape input.

The minimal universal Turing machine the author talks about is [1].

[1] https://en.wikipedia.org/wiki/Wolfram%27s_2-state_3-symbol_T...


I once floated the idea of a "simplest Turing Machine" contest that limited materials to wood, nails, screws, and string. The one with the least number of parts wins. Each Nail and screw would be considered half a part, but total score would be rounded up. This is because people tend to consider fasteners as ancillary, but not counting them at all would create too many loopholes and oddities.


A wonderful piece of work!

But this remark in the accompanying document did make me laugh:

> Another search on the internet yielded a small list of machines built by others. Several had electronic controllers controlling the mechanics. I can hear Mr. Turing say, “A computer running a computer. Whaaaat?”

I'm not sure Dr Turing would have expressed surprise in that manner, and even less sure that he would have been surprised that one computing machine could simulate the behaviour of another.


My first question is, could this have been made at the time Babbage was trying to build the Analytical Engine, which (had he completed it) would have been the first Turing-complete machine? Would it have accomplished the same practical results (at least with some scale-up)?

https://en.wikipedia.org/wiki/Analytical_Engine


Personal speculation:

There's nothing about this that someone in Babbage's time could not have build. Removing the negatives from that statement ... this could easily have been built in Babbage's time.

However, it would not have been. A literal Turing machine needs a huge tape and a very, very long time to be able to do useful calculations. Babbage's Difference Engine could be seen to be immediately useful performing a task that was essential at the time. Babbage's proposed Analytical Engine was less obviously immediately useful, but it could do things that were obviously on the way to be useful, so might have got some speculative funding.

But a naked, vanilla Turing Machine is so obviously not (directly) useful, and no one of the time would having funding building one that was big enough to do anything of practical interest.


Indeed. The 'Turing machine' was proposed by Turing as a thought experiment. It was the simplest implentation of a universal computing machine that he could think of. Any physical implementation of it would be incredibly inefficient.


Well, one could invent some mean VM to run upon a universal TM, and program it instead. With some hardware codec to translate inputs/outputs between this VM and the UTM... but yeah, if you want to do arithmetics, start with Odhner pinwheels.


Your question reminds me of a russian humorous story somewhere from the 90s or early 00s. Called "Wooden Computational Machine", it describes an wooden computer created for the russian zar and the chaos that ensues. The author is called Maxim Samohvalov, but I couldn't find an english translation. If someone can find such, please, share.


A Turing Machine can be a very simple device: it depends on the size of the transition function. The model in the video has a small one. Implementing the Analytical Engine would make it pretty big.


Suppose someone two thousand years ago saw the plans for the Turing machine in the video. How might that have shaped history? Whether it was practical or not is not really the question, but rather how might it shape future technologies? Makes me think of da Vinci's notebook.


Forget nanobots; I believe that a "clanking" self-replicating machine can be made from sources of wood and metal feedstock (and mechanical power). This could be the control unit.

Is anyone else interested in this kind of thing? Looking for community.


L.S. and S.V. Penrose did some experiments around this: https://www.youtube.com/watch?v=2_9ohFWR0Vs


Thanks for the link, that's interesting.

All the "self-replicators" I've come across are either digital (e.g. Conway's Game of Life) or like the ones in this video, that is they require specialized parts as their "food." You could say they are all "carnivores" in the sense that they consume other machines. My interest is in "herbivore" self-replicators, those that reproduce by consuming raw materials and energy from their environment.


The cell does some mind boggling stuff. What the macro organism achieves is boring in comparison. Hence clanking replication could be very hard.


Agreed; I'm proud of our little constituent buddies!

I can't help thinking about how close a water-powered lumber mill feels to full closure (i.e. it produces all the parts it uses). Humans perform the chopping, move things from station to station, and assemble the finished parts into new buildings and machines.


What about the saw blade? It's probably hard to cut wood with wood...

Usually you require a few different material "paths" for replication. You have a hard material that can be softened with chemicals for example, so it can be shaped with machinery that is softer than the end product.


Somehow I expect there to exist a minimal mechanical Turing machine, I expect it to be highly symmetrical and I expect it to be very beautiful and fascinating to watch.


So could one enter the code of an universal turing machine here and write a program directly on the tape?


Being able to take the configuration "table" and change the input quickly is a neat feature!!


Too bad they didn't give it an infinite wooden tape :)


I cheated by letting it sit around a long time, and termites ended up doing all the work.


Want! :)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: