@@Mr.Carrot an and gate is just three linked not gates (and very theoretically speaking an or gate in between) i think the real challenge is making an or gate
@@chr1ss99 Not really, an or gate can be built easily from and and not: not (not x and not y) = x or y because of de Morgan's laws. And now with and, or and not gates every possible truth table can trivially be built. There's probably a solution for or involving less scissors though Edit: I missed the word "table"
@@jjtt Well yes, an and gate can be build from and and not, but an and gate requires an or gate (when the two not'ed scissors merge, to be not'ed again) so that solution doesn't work im practically just saying, that it will be hard to merge the truths of two scissors in a single connection
My first thought with this gate is that it might be possible to simplify it and make it smaller. What if it wasn't a scissor computer but just a lever computer with each lever fixed in the center. A lever pointing up could be "true" and down would be "false". Then your not gate would be a lever but idk where the second input would be for the and gate...
I think the non-constant length is actually a bonus, because of how it allows you to switch it by moving the ends closer and further, and i'd want to build in a way to exploit that. edit: I can think of how to make an XOR gate, and I think I can make an AND gate out of a combination of the XOR and NOT gates... hmm...
The easiest way to do that would be to have fewer links to one side of the 90° linkage pair. Giving the left side 1 more link than the right makes the difference in overall length between the left side being open and closed equivalent to the difference in overall length of 1 regular scissor (edit: maybe not exactly, but nearly). If we'd bother to make the linkage have positive difference like this, we can do better, because we probably don't have monotonic difference in length with respect to the angle. I presume just adding more regular linkages to the same side will eventually overpower the component that makes it not monotonic.
unfortunately, I don't think XOR and NOT gates shouldn't be able to make AND (or OR). One way to see it is because both have "even" truth tables. But maybe there's a direct way to build a scissor OR or AND or NAND?
@@user-sl6gn1ss8p you're thinking too inside the box. you can use the physicality of these gates to do things that an electrical logic gate couldn't. my solution isn't very elegant, I would like a more direct way, but I'll probably make a video soon demonstrating how it's possible
@@PeterBarnes2 that wouldn't work by simmetry, remember it can be on or off for both sides so if one compensates for the other when it's in one state it'll mess up worse the other state.
Solution for the length problem: add a component to the output that just takes the vertical component and lets the horizontal slide. For example, a vertical groove is fixed in place, then put two bars in it that are forced to be at right angles. Inside of those two horizontal bars, put a groove that lets the ends of the scissors slide freely along them, but the vertical motion will translate into the right side of those bars.
This is an unbelievably powerful "cheeky" answer, which I love. To add to this, if you think about it, you only need the 2 L pieces for the NOT gate, which might actually end up making the bulk of the simplest scissor gate be the correcting grooves, and it would still feel quite "scissor-y" in its mechanical complexity.
I wonder if you might get some moments of strong stresses in weird places as certain lengths force strong pressure on the bars to push themselves up when they want to move out. E.g. 85 degrees into the bar you have to exert a lot of force into the bar to push that piece up. This force would transfer back into the mechanism and start stressing the joints. Compound that with needing to push potentially thousands of other gates entirely through this mechanism and I don't think this machine will be turing complete any time soon.
Mechanical stress could be overcome by some kind of active buffering or decoupling mechanism to compensate for friction in the system and propagation delays due to slop.
*Solution for AND gate,* using three *NOT* gates. Sorry, no monospaced fixed width fonts, emojis shall be used. Better viewed in horizontal orientation in small screen devices. *Emoji Notation:* ↕️ Open input/output, Zero/False 0️⃣ ↔️ Close input/output, One/True 1️⃣ 🔃 Open link (high height); 🔁 Closed link (low height); ⬜ Emoji space alignment (only, no effect); ⏫⏬ Fixed points: top and bottom; ↗️↘️ Additional links: top and bottom, to give some space to the scissors; 🔀 *Overlap solution:* two additional links that are *NOT* connected; *AND operation:* (there is a vertical movement of last scissor during operation) 0️⃣⬜⬜⏫ ↕️🔃🔁↘️⬜⬜⬜0️⃣ ⬜⬜⬜⬜🔀🔁🔃↕️ ↕️🔃🔁↗️ 0️⃣⬜⬜⏬ 0️⃣⬜⬜⏫⬜⬜⬜0️⃣ ↕️🔃🔁➡️🔀🔁🔃↕️ ⬜⬜⬜⬜↗️ ↔️🔁🔃↗️ 1️⃣⬜⬜⏬ 1️⃣⬜⬜⏫ ↔️🔁🔃↘️ ⬜⬜⬜⬜↘️⬜⬜0️⃣ ↕️🔃🔁➡️🔀🔁🔃↕️ 0️⃣⬜⬜⏬ 1️⃣⬜⬜⏫ ↔️🔁🔃↘️⬜⬜⬜1️⃣ ⬜⬜⬜⬜🔀🔃🔁↔️ ↔️🔁🔃↗️ 1️⃣⬜⬜⏬ For non-extensible _"buffer"_ (0 to 0, 1 to 1) use two *NOT* gates in series: 0️⃣⬜⬜1️⃣⬜⬜0️⃣ ↕️🔃🔁↔️🔁🔃↕️ 1️⃣⬜⬜0️⃣⬜⬜1️⃣ ↔️🔁🔃↕️🔃🔁↔️ I hope this helps.
Maybe this revision is better for understanding. *AND operation:* (there is a vertical movement of the last scissor during operation). 🔀 *Overlap links* crossed but *not* connected - perhaps this is the best way to emulate a "switching" *AND* gate. 0️⃣⬜⬜➡️⏫ ↕️🔃🔁 ⬜⬜⬜➡️↘️⬜⬜⬜⬜0️⃣ ⬜⬜⬜⬜⬜🔀↔️🔁🔃↕️ ⬜⬜⬜➡️↗️ ↕️🔃🔁 0️⃣⬜⬜➡️⏬ 0️⃣⬜⬜➡️⏫ ↕️🔃🔁 ⬜⬜⬜➡️↘️⬜⬜⬜⬜0️⃣ ⬜⬜⬜⬜⬜🔀↔️🔁🔃↕️ ⬜⬜⬜⬜↗️ ⬜⬜⬜↗️ ↔️🔁🔃 1️⃣⬜⬜➡️⏬ 1️⃣⬜⬜➡️⏫ ↕️🔃🔁 ⬜⬜⬜↘️ ⬜⬜⬜⬜↘️⬜⬜⬜⬜0️⃣ ⬜⬜⬜⬜⬜🔀↔️🔁🔃↕️ ⬜⬜⬜➡️↗️ ↕️🔃🔁 0️⃣⬜⬜➡️⏬ 1️⃣⬜⬜➡️⏫ ↕️🔃🔁 ⬜⬜⬜↘️⬜↗️⬜⬜⬜1️⃣ ⬜⬜⬜⬜🔀⬜↕️🔃🔁↔️ ⬜⬜⬜↗️⬜↘️ ↔️🔁🔃 1️⃣⬜⬜➡️⏬
Note that the crossing of the overlapping links retreats in the direction of the inputs (left) when both are in state "1" to open (zero) the input of the last NOT scissors.
*OR operation:* (with some vertical movement of the last scissor). No overlapping links and the fixed point are internal. Maybe external additional links are necessary to prevent "half-opening" when one of the inputs is "1" and the other is "0". 0️⃣ ↕️🔃🔁↘️ ⬜⬜⬜⏬⬇️⬜⬜0️⃣ ⬜⬜⬜⬜↔️🔁🔃↕️ ⬜⬜⬜⏫⬆️ ↕️🔃🔁↗️ 0️⃣ 0️⃣ ↕️🔃🔁↘️ ⬜⬜⬜⏬⬇️⬜⬜1️⃣ ⬜⬜⬜⬜↕️🔃🔁↔️ ⬜⬜⬜⏫⬆️ ↔️🔁🔃⬜⬆️ 1️⃣⬜⬜➡️ 1️⃣⬜⬜➡️ ↔️🔁🔃⬜↘️ ⬜⬜⬜⏬⬇️⬜⬜1️⃣ ⬜⬜⬜⬜↕️🔃🔁↔️ ⬜⬜⬜⏫⬆️ ↕️🔃🔁↗️ 0️⃣ 1️⃣⬜⬜➡️ ↔️🔁🔃⬜↘️ ⬜⬜⬜⏬⬇️⬜⬜1️⃣ ⬜⬜⬜⬜↕️🔃🔁↔️ ⬜⬜⬜⏫⬆️ ↔️🔁🔃⬜↗️ 1️⃣⬜⬜➡️
*Flip-flop:* one-bit memory. *SET* is *1* = *M* output goes to 1 to keep this value even when *SET* becomes back to *0* . *RESET* is *1* = memory is cleared as *M* goes to *0* to keep this value even when *RESET* becomes back to *0* . Inverse to electronic Boolean logic, which only has scalar values (volts), Mechanics has vector values that mean it has direction. This means that the "open" movement has an opposite direction of the "close" movement. So two *NOT* gates are required to feed the memory output back to the flip-flop input - they keep the scalar value of boolean logic and compensate the vectorial direction. These right-to-left operation gates would not be needed in a digital electronic circuit. The *OR-AND* latch was chosen because there is only one feedback link ( *right-to-left operation* - see the direction of the arrows), and *NOT* gates are available. The *NOT* gate scissors required for 1-bit memory are 3 *NOT* gates for *OR* gate, 3 *NOT* gates for *AND* gate, 1 *NOT* gate for *RESET* input, and 2 *NOT* gates for feedback link. Nine *NOT* gate scissors in total. *Emoji notation:* 🆖▶️⏺️ - *NOT* gate - *NG>o* ; 🅾️ *OR* gate - *[O]* ; 🅰️ *AND* gate - *[A]* ; 🆚 *SET* input - *[VS]* ; ®️ *RESET* input - *(R)* ;
One thing to consider is that you can have two-input two-output gates. Despite sounding like it has more pieces, this is often easier, and many gates are built this way and just don't connect one of the two outputs. Usually the two outputs are inverses of each other, Q and ~Q. If you built an AND gate this way you would also get a NAND output. Now, whether this applies to scissor logic, I don't know.
I was obsessed with this scissor mechanism as a kid. All I wanted was a handheld scissor thing so I could reach things that are far away. I'd picture it like a reacher-grabber that could reach-grab really far and it was just such an insanely cool concept to me. Honestly it still is. Not only this but I love fidgeting with pretty much anything. So, you can imagine my reaction, with these past experiences, and upon seeing what you did at 1:25 it made me want one so very badly. And reminded me of how badly I wanted such a scissor-based contraption when I was younger. Thank you for this video.
The nice thing about using a scissor mechanism is that assuming that you don't have any slack in the mechanism, setting the output will give back the input that would've caused it. This means that you end up with reversible logic gates like the Toffoli/CCNOT gate, which also have to be bijective (seeing as any overdetermination requires slack in the mechanism) and that anything you build out of scissors without slack will be reversible - so a multiplier is inherently a divider etc. Of course this means that things like ordinary AND and XOR gates are completely out of the picture (in the slackless world) but CCNOT Universality is a known property in reversible computing, so you should still be able to encode arbitrary boolean functions, with the added benefit that they will be reversible but overspecified, ie. more output bits than is needed to encode the actual output of the encoded boolean function.
For the NOT gate, I could use it to make alternating billows for my forge that run on a single crank shaft. This would allow a more even flow of air and a more consistent heat.
You can make an (almost) constant length normal gate (the one who conserves the logic state) by connecting two NOT Gates in series, since the latter already almost conserves length.
About AND gate and OR gate: AND can be explained as "min" function. Likewise OR is a "max" function. My suggestion: make two input scissors on top of each other with the third scissors as an output. 1. The OR gate: the input scissors will try to widen the output (push from the inside). Add a spring that tries to narrow the output points back. This way each of the inputs will independently widen the output (make it true), but if both inputs are false - the spring will narrow the output down (make it false). Same can be done to the AND gate, but here the inputs will try to narrow the output and the spring (or two from the sides) will try to widen it back. Hope the creator will eventually see this....
The change in length is primarily due to the centre portion - replacing the right-angles with 60 degree angles, using an equilateral triangle the base shape as opposed to a square, would fix the change in length at each extreme, and the middle, of the lengthwise motion. I'm unsure of the full length profile through the points between these positions, however, and the changes to the rest of the mechanism to compensate would also reduce the range of crosswise motion between each pair of opposing points. A tricky one, to be sure!
With a bit of angle chasing and a little trigonometry one could conjure up a function that spits out the total length of this contraption based on the input angle. Then it's only a matter of checking if it's constant or not
All these comments talking about computers and technical stuff, then there's me not understanding any of that and thinking "This looks like a fun thing to fidget with"
I feel like if you wanted more mechanically interesting movements then I would add additional layers. So the same scissoring motion would produce different movements on different layers. I feel like that would lead to more possibility for combos and computations. Could possibly make for a fun calculator where you flip some switches and then pull a big lever raising the scissors up to compute the answer.
The harder question is, can you make every gate, and just a wire, with constant length, or else your entire system will break when something changes length…
I'm thinking if we do make it, we'd probably need a huge space to do it first, like how early computers were vs how they are now (entire basketball courts vs your typical smart phone).
I think one can create an XOR gate by using 4 of those NOT scissors. First, place 2 of those NOT scissors parallel to each other. Then, connect 1 NOT scissors perpendicular to them at one end. Now it would look like a コ shape. This way, it would shift shapes left or right only when one input is True, but jams at the middle when both are True. Finally, attach a short stick to that sideways NOT scissors' middle point, and place it in between the 4th NOT scissors' arms. It will open when the perpendicular one shift shapes. Attach a small circle at the pivot point of the 4th NOT scissors' arms. This makes sure the stick pushes the 4th NOT scissors back in place once it's opened.
@@Owen-bk5fc I haven't thought about what would happen when the inputs aren't changed simultaneously. That might happen if both inputs' arms were directly connected to the perpendicular one. I'm still thinking about a solution to isolate them without over complicating it with strings and rubber bands and stuff.
@@Owen-bk5fc Currently, the (00) and (11) state works by balancing the opposing forces from the two inputs. So if you pinch one of the inputs halfway through, it can change into that (00)/(11) state you expect; but if you pinch too hard then it modifies the other input too.
Going on pure instinct, for the constant length; two ideas figuring them out as I type. The movement comes from the rotation of the connection line between the axes: it's a square pivoting on a corner with the axes on the connected corners You need to get the overall rotation point in between the axes. But with a fixed axis that won't work. So: have that move along a curved path, sliding along a shaped "linear" axis. Or Extending the length of the first/last joints changes how agressively it extends. It feels like that should help somehow. There's lots of "interesting mechanical mechanisms" demonstrated on here, many dedicated to hinging arms (like on the channel tenfold). there's probably one that is applicable here? What is also interesting is that you also have a kind of bit here: left or right, somewhat fixed. That's a 0/1
I'd like to make note that a one link scissor is a Not gate just shifted by 45 degrees. This is because as one opens the scissor it closes the sides. This should help simplify any scissor circuit.
its interesting people find these logic mechanisms everywhere, why nobody ever built anything out of them is reliability, even in the days of analogue computers, they tried everything but cams, gears and rods were the most reliable versions that could run indefinitely
Hey Henry - long time, no see. Not done the analysis, but taking a lead from electrical theory, three phase might work for lengths - i.e. two sections of 120 degrees scissors, with a longer middle section. Trouble is, that’d destroy the not and make it a “buffer” again - like the top one - but one of constant length. I’ve been idly mulling the “and” issue - nothing truly elegant is presenting itself. I hope that gets solved - it’d be nice to see even a full adder. I suspect friction will start to be an issue!
Just figured out how to make simple OR/AND gates : Arrange 2 NOT scissors and 1 regular scissors into an 工 shape. Both NOT scissors' middle points connect to one end of the regular scissors. Add a third NOT scissors and form a 王 shape, its middle point connecting to the regular scissors' one. Now depending on which direction you shift the third one's position, you can make either an OR gate or an AND gate. Notice when one of the inputs is True, it stretches the regular scissors and forms a Z shape. By moving the third NOT scissors slightly to the left or right, you're essentially flooring or ceiling that 0.5 state into a 0 or 1.
Fun thing about a not gate. If you put it's output back into it's input it oscillates. I imagine because this mechanism has a neutral state it would just be stuck at neutral.
Just found this video. Liked it enough to subscribe after just one video. Very interesting idea. I would love to see a scissor link computer. (Also had to like and comment for the TH-cam engagement)
Some kind of clutch would be a useful building block for an AND gate or a CNOT gate. One input acts as a control to allow or prevent propagation of signals in one or more target linkages. I could imagine this working by coupling the target input and output by way of a broken pair of straight linkages (not scissored). The control input is mounted orthogonally to the target and acts to either hold the broken linkages straight and parallel to one another (allowing the signal to pass) or by letting them buckle in the middle (preventing transfer of the signal). A single control input could link control multiple targets simultaneously which would be useful for multiplexing. To make an AND out of a clutch you can either add a spring to provide some restoring force in the target output when the clutch is disengaged or combine a complementary pair of controlled gates in some other way.
That 90 degree piece has a name, it's a bell crank. I highly recommend you check out @ArglinV3 and @thang010146 for linkages that perform logical operations.
If you want to create a "pass" node, that outputs its input value (basically a wire) in a similar manner, you should connect two (or any even number) NOTs, and assuming the constant length problem is solved, these wires would also be constant length. Also they would be really cool visually, blending in in a larger machine. This would also have greater mechanical strength than some of the other ideas I have for just wires, and these can actually be suspended in any direction. Another huge problem with the future designs is buffering. You can manage to create logical blocks that perform a few operations as gates, but after a certain complexity you either need very good bearings, which is an expensive option (and still not infinitely scalable), or some kind of logical buffers, that "amplify the signal" passed to it. This is a pretty straightforward electromechanical challenge, but the whole point is to only use this reduced set of building blocks. I'm thinking maybe something with "rechargeable" torsion springs and latches like in a mechanical clock.
'These two look like they're identical mechanisms' Me: *Proceeds to pause the video and find the difference between the two* (and found the difference)
some notes: - i think the grooves idea like below is a nice way to make the length constant - it seems like you'd more or less be limited to trigonometry because of how scissor linkages work. and you can't really make them add up to a constant. there's the good old pythagorean theorem but that requires sin^2 and cos^2 so that doesn't work either - you *may* be able to get away with having some curved linkages - potentially just at the ends, potentially differently curved for each half (plus curved L pieces - if each half is differently curved you may (or may not) want the L pieces to be asymmetric) - for 2-input gates you'd probably want some kind of junction. either coming in from three sides of a square, or three sides of a triangle - for convenience (well, to be able to make complex circuits) you would likely also want a simple mechanism to rotate a truth value by say, 90 or 120 degrees - (i'm not sure how logic gates would work - you may need additional degrees of freedom like the grooves?)
For and gate, two not gates beside each other with the inner outputs linked with the identity linkage? (Outer ends of the not gates fixed I suppose) It wouldn’t be perfect in terms of deciding power - both off would be markedly different from one on would be markedly different from two on - but it should be closer maybe?
I'm not a math guy, but I think you can also create a signal inverter by introducing an orthogonal junction in the links - as you expand one dimension, you contract the other. You might be able to create a parallel inverter with the use of this pseudo-NOT gate's constant length. You could create an absolute NOT gate by eliminating the second half of the links to terminate the extension throughput.
I would reframe the whole mechanic a bit: suppose one arm of the input is fixed to a grid, then you only look at the movement of the other arm. This reduces the output to one vertical movement too. Now you could imagine a rectangular frame that can only move vertically. The horizontal movement of the scissors along the horizontal part of the rectangle is ignored, you just have some leeway for the length change there. The vertical movement is the output you are interested, you get that from only one point of the two output scissors, the other one is moving in a horizontal rail to give it leeway for the length change, but it is fixed in the vertical axis. Those frames are the "wires" that you use to connect to the next stage of logic gates. It's not a fully scissor link mechanism this way, but building a first prototype with those moving frames could give you some insight in how a logic circuit could work. Maybe you find a scissor variation of that idea later.
If we're aiming to not do anything too cheeky (like a corrector segment at the end), it seems to me like your inconsistent length is caused by 6x the hypotenuse of a right angle triangle made by the line segments being longer than 6x the length of the line segments. The obvious conceptual answer to this is to somehow change things so that we're measuring equilateral triangles instead, but I can't for the life of me picture what that would look like.
if you can measure the distance between the arms on each side, and hook it up to an osilating mechanism, you can make an encoder for counting whatever causes the osilations.
You could extract two bits of information ( ie. Light which goes through the scissors and the length variable for each scissor. Although the light would be position dependent for the second scissor.
People really be using everything to make logic gates, except of transistors. I’ve seen it with water, fireworks, RC cars, many other mechanical constructions and now even with scissors. 😂
I wonder how stable it could be at the center L joint. Since it’s anchored at both ends, and can be driven from something stationary. Could be an interesting choice for something like an extrusion 3D printer. Or maybe since it converts a very long motion into a comparatively fine one it could be a tool for firmly but precisely gripping fine objects.
You say scissors based mechanical computer like it's some kind of joke, but there is reason to believe that the cytoskeleton(a cross linkage of various proteins that support the form, movement, and metabolism of any cell), besides simply acting as scaffolding, is actually encoding information and performing computations which direct the cell's metabolism and behavior. This can explain how a single-celled organism can have fairly complex behavior, but also how cells organize themselves in general. I bet that the some of the proteins that make up the cytoskeleton can have distinct configuration states, some which are just like the scissor arm which can be elongated or compressed, and there is a some kind of junction protein which acts similarly to your not gate.
In order to make one of (Very close to) constant length you would need a link that is at the middle point when the input is open and open when the input is closed.
I was immediately thinking that going through 3rd dimension is a quick lazy answer. Add hinges to the arms so ends of the mechanism stay a fixed (arbitrarily adjustable) distance from each other but the middle of it bows a up and down, freely, in sort of an arch or wrinkled shape.
I like that this allows you to output true to true and false to false without changing the overall length (too much) as compared to a "normal" scissor linkage. Not sure of a practical use, but it is interesting :)
i remember you could do logic with just or, not and falsum. Or rhe duals of course. I have looked at a TH-cam search and there seems be a LEGO Technic implementation
Here before someone builds a computer that adds two 4 bit integers made entirely out of scissors
They have to design an and gate first
@@Mr.Carrot an and gate is just three linked not gates (and very theoretically speaking an or gate in between)
i think the real challenge is making an or gate
@@chr1ss99 Not really, an or gate can be built easily from and and not: not (not x and not y) = x or y because of de Morgan's laws. And now with and, or and not gates every possible truth table can trivially be built. There's probably a solution for or involving less scissors though
Edit: I missed the word "table"
@@jjtt Well yes, an and gate can be build from and and not, but an and gate requires an or gate (when the two not'ed scissors merge, to be not'ed again) so that solution doesn't work
im practically just saying, that it will be hard to merge the truths of two scissors in a single connection
@@jjtt So our and gate contains an or gate, and our or gate contains an and gate? I'm beginning to see a problem...
FINALLY one step closer to a "scissors-based mechanical computer". I love your videos, Henry.
In 6 billion years when we've used up all the silicon on earth for non-scissors-based computing, your comment is going to look pretty short sighted. 😅
ha
My first thought with this gate is that it might be possible to simplify it and make it smaller.
What if it wasn't a scissor computer but just a lever computer with each lever fixed in the center. A lever pointing up could be "true" and down would be "false". Then your not gate would be a lever but idk where the second input would be for the and gate...
I think the non-constant length is actually a bonus, because of how it allows you to switch it by moving the ends closer and further, and i'd want to build in a way to exploit that.
edit: I can think of how to make an XOR gate, and I think I can make an AND gate out of a combination of the XOR and NOT gates... hmm...
The easiest way to do that would be to have fewer links to one side of the 90° linkage pair. Giving the left side 1 more link than the right makes the difference in overall length between the left side being open and closed equivalent to the difference in overall length of 1 regular scissor (edit: maybe not exactly, but nearly).
If we'd bother to make the linkage have positive difference like this, we can do better, because we probably don't have monotonic difference in length with respect to the angle. I presume just adding more regular linkages to the same side will eventually overpower the component that makes it not monotonic.
unfortunately, I don't think XOR and NOT gates shouldn't be able to make AND (or OR). One way to see it is because both have "even" truth tables.
But maybe there's a direct way to build a scissor OR or AND or NAND?
@@user-sl6gn1ss8p you're thinking too inside the box. you can use the physicality of these gates to do things that an electrical logic gate couldn't.
my solution isn't very elegant, I would like a more direct way, but I'll probably make a video soon demonstrating how it's possible
Could you please guys check my comment if the AND with three NOT gates could work? thanks
@@PeterBarnes2 that wouldn't work by simmetry, remember it can be on or off for both sides so if one compensates for the other when it's in one state it'll mess up worse the other state.
Solution for the length problem: add a component to the output that just takes the vertical component and lets the horizontal slide. For example, a vertical groove is fixed in place, then put two bars in it that are forced to be at right angles. Inside of those two horizontal bars, put a groove that lets the ends of the scissors slide freely along them, but the vertical motion will translate into the right side of those bars.
This is an unbelievably powerful "cheeky" answer, which I love. To add to this, if you think about it, you only need the 2 L pieces for the NOT gate, which might actually end up making the bulk of the simplest scissor gate be the correcting grooves, and it would still feel quite "scissor-y" in its mechanical complexity.
I wonder if you might get some moments of strong stresses in weird places as certain lengths force strong pressure on the bars to push themselves up when they want to move out. E.g. 85 degrees into the bar you have to exert a lot of force into the bar to push that piece up. This force would transfer back into the mechanism and start stressing the joints. Compound that with needing to push potentially thousands of other gates entirely through this mechanism and I don't think this machine will be turing complete any time soon.
Mechanical stress could be overcome by some kind of active buffering or decoupling mechanism to compensate for friction in the system and propagation delays due to slop.
Could you please guys check my comment if the AND with three NOT gates could work? thanks
you could also just use multiple not gates, and it will stay the same length
for example, >||< (| indicates not gate)
*Solution for AND gate,* using three *NOT* gates.
Sorry, no monospaced fixed width fonts, emojis shall be used. Better viewed in horizontal orientation in small screen devices.
*Emoji Notation:*
↕️ Open input/output, Zero/False 0️⃣
↔️ Close input/output, One/True 1️⃣
🔃 Open link (high height);
🔁 Closed link (low height);
⬜ Emoji space alignment (only, no effect);
⏫⏬ Fixed points: top and bottom;
↗️↘️ Additional links: top and bottom, to give some space to the scissors;
🔀 *Overlap solution:* two additional links that are *NOT* connected;
*AND operation:* (there is a vertical movement of last scissor during operation)
0️⃣⬜⬜⏫
↕️🔃🔁↘️⬜⬜⬜0️⃣
⬜⬜⬜⬜🔀🔁🔃↕️
↕️🔃🔁↗️
0️⃣⬜⬜⏬
0️⃣⬜⬜⏫⬜⬜⬜0️⃣
↕️🔃🔁➡️🔀🔁🔃↕️
⬜⬜⬜⬜↗️
↔️🔁🔃↗️
1️⃣⬜⬜⏬
1️⃣⬜⬜⏫
↔️🔁🔃↘️
⬜⬜⬜⬜↘️⬜⬜0️⃣
↕️🔃🔁➡️🔀🔁🔃↕️
0️⃣⬜⬜⏬
1️⃣⬜⬜⏫
↔️🔁🔃↘️⬜⬜⬜1️⃣
⬜⬜⬜⬜🔀🔃🔁↔️
↔️🔁🔃↗️
1️⃣⬜⬜⏬
For non-extensible _"buffer"_ (0 to 0, 1 to 1) use two *NOT* gates in series:
0️⃣⬜⬜1️⃣⬜⬜0️⃣
↕️🔃🔁↔️🔁🔃↕️
1️⃣⬜⬜0️⃣⬜⬜1️⃣
↔️🔁🔃↕️🔃🔁↔️
I hope this helps.
Amazing.
Maybe this revision is better for understanding.
*AND operation:* (there is a vertical movement of the last scissor during operation).
🔀 *Overlap links* crossed but *not* connected - perhaps this is the best way to emulate a "switching" *AND* gate.
0️⃣⬜⬜➡️⏫
↕️🔃🔁
⬜⬜⬜➡️↘️⬜⬜⬜⬜0️⃣
⬜⬜⬜⬜⬜🔀↔️🔁🔃↕️
⬜⬜⬜➡️↗️
↕️🔃🔁
0️⃣⬜⬜➡️⏬
0️⃣⬜⬜➡️⏫
↕️🔃🔁
⬜⬜⬜➡️↘️⬜⬜⬜⬜0️⃣
⬜⬜⬜⬜⬜🔀↔️🔁🔃↕️
⬜⬜⬜⬜↗️
⬜⬜⬜↗️
↔️🔁🔃
1️⃣⬜⬜➡️⏬
1️⃣⬜⬜➡️⏫
↕️🔃🔁
⬜⬜⬜↘️
⬜⬜⬜⬜↘️⬜⬜⬜⬜0️⃣
⬜⬜⬜⬜⬜🔀↔️🔁🔃↕️
⬜⬜⬜➡️↗️
↕️🔃🔁
0️⃣⬜⬜➡️⏬
1️⃣⬜⬜➡️⏫
↕️🔃🔁
⬜⬜⬜↘️⬜↗️⬜⬜⬜1️⃣
⬜⬜⬜⬜🔀⬜↕️🔃🔁↔️
⬜⬜⬜↗️⬜↘️
↔️🔁🔃
1️⃣⬜⬜➡️⏬
Note that the crossing of the overlapping links retreats in the direction of the inputs (left) when both are in state "1" to open (zero) the input of the last NOT scissors.
*OR operation:* (with some vertical movement of the last scissor).
No overlapping links and the fixed point are internal. Maybe external additional links are necessary to prevent "half-opening" when one of the inputs is "1" and the other is "0".
0️⃣
↕️🔃🔁↘️
⬜⬜⬜⏬⬇️⬜⬜0️⃣
⬜⬜⬜⬜↔️🔁🔃↕️
⬜⬜⬜⏫⬆️
↕️🔃🔁↗️
0️⃣
0️⃣
↕️🔃🔁↘️
⬜⬜⬜⏬⬇️⬜⬜1️⃣
⬜⬜⬜⬜↕️🔃🔁↔️
⬜⬜⬜⏫⬆️
↔️🔁🔃⬜⬆️
1️⃣⬜⬜➡️
1️⃣⬜⬜➡️
↔️🔁🔃⬜↘️
⬜⬜⬜⏬⬇️⬜⬜1️⃣
⬜⬜⬜⬜↕️🔃🔁↔️
⬜⬜⬜⏫⬆️
↕️🔃🔁↗️
0️⃣
1️⃣⬜⬜➡️
↔️🔁🔃⬜↘️
⬜⬜⬜⏬⬇️⬜⬜1️⃣
⬜⬜⬜⬜↕️🔃🔁↔️
⬜⬜⬜⏫⬆️
↔️🔁🔃⬜↗️
1️⃣⬜⬜➡️
*Flip-flop:* one-bit memory.
*SET* is *1* = *M* output goes to 1 to keep this value even when *SET* becomes back to *0* .
*RESET* is *1* = memory is cleared as *M* goes to *0* to keep this value even when *RESET* becomes back to *0* .
Inverse to electronic Boolean logic, which only has scalar values (volts), Mechanics has vector values that mean it has direction. This means that the "open" movement has an opposite direction of the "close" movement. So two *NOT* gates are required to feed the memory output back to the flip-flop input - they keep the scalar value of boolean logic and compensate the vectorial direction. These right-to-left operation gates would not be needed in a digital electronic circuit.
The *OR-AND* latch was chosen because there is only one feedback link ( *right-to-left operation* - see the direction of the arrows), and *NOT* gates are available. The *NOT* gate scissors required for 1-bit memory are 3 *NOT* gates for *OR* gate, 3 *NOT* gates for *AND* gate, 1 *NOT* gate for *RESET* input, and 2 *NOT* gates for feedback link. Nine *NOT* gate scissors in total.
*Emoji notation:*
🆖▶️⏺️ - *NOT* gate - *NG>o* ;
🅾️ *OR* gate - *[O]* ;
🅰️ *AND* gate - *[A]* ;
🆚 *SET* input - *[VS]* ;
®️ *RESET* input - *(R)* ;
One thing to consider is that you can have two-input two-output gates. Despite sounding like it has more pieces, this is often easier, and many gates are built this way and just don't connect one of the two outputs. Usually the two outputs are inverses of each other, Q and ~Q. If you built an AND gate this way you would also get a NAND output.
Now, whether this applies to scissor logic, I don't know.
Ruin with it!
I was obsessed with this scissor mechanism as a kid. All I wanted was a handheld scissor thing so I could reach things that are far away. I'd picture it like a reacher-grabber that could reach-grab really far and it was just such an insanely cool concept to me. Honestly it still is. Not only this but I love fidgeting with pretty much anything. So, you can imagine my reaction, with these past experiences, and upon seeing what you did at 1:25 it made me want one so very badly. And reminded me of how badly I wanted such a scissor-based contraption when I was younger. Thank you for this video.
if only you had an ultra hand
The nice thing about using a scissor mechanism is that assuming that you don't have any slack in the mechanism, setting the output will give back the input that would've caused it. This means that you end up with reversible logic gates like the Toffoli/CCNOT gate, which also have to be bijective (seeing as any overdetermination requires slack in the mechanism) and that anything you build out of scissors without slack will be reversible - so a multiplier is inherently a divider etc.
Of course this means that things like ordinary AND and XOR gates are completely out of the picture (in the slackless world) but CCNOT Universality is a known property in reversible computing, so you should still be able to encode arbitrary boolean functions, with the added benefit that they will be reversible but overspecified, ie. more output bits than is needed to encode the actual output of the encoded boolean function.
Not only is it a NOT gate, but under slight compression it can also be a flip flop of sorts. Excellent.
For the NOT gate, I could use it to make alternating billows for my forge that run on a single crank shaft. This would allow a more even flow of air and a more consistent heat.
You can make an (almost) constant length normal gate (the one who conserves the logic state) by connecting two NOT Gates in series, since the latter already almost conserves length.
About AND gate and OR gate:
AND can be explained as "min" function.
Likewise OR is a "max" function.
My suggestion: make two input scissors on top of each other with the third scissors as an output.
1. The OR gate: the input scissors will try to widen the output (push from the inside). Add a spring that tries to narrow the output points back. This way each of the inputs will independently widen the output (make it true), but if both inputs are false - the spring will narrow the output down (make it false).
Same can be done to the AND gate, but here the inputs will try to narrow the output and the spring (or two from the sides) will try to widen it back.
Hope the creator will eventually see this....
I wish I had the time to investigate all this neat stuff. I love this guy's videos. Super cool. great channel
It's so satisfying seeing the bottom one "self-balance".
Why am I obsessed with auto-balancing mechanics?
before you operated the mechanism i paused the video and closely looked at them. pretty neat mechanism.
The change in length is primarily due to the centre portion - replacing the right-angles with 60 degree angles, using an equilateral triangle the base shape as opposed to a square, would fix the change in length at each extreme, and the middle, of the lengthwise motion. I'm unsure of the full length profile through the points between these positions, however, and the changes to the rest of the mechanism to compensate would also reduce the range of crosswise motion between each pair of opposing points.
A tricky one, to be sure!
With a bit of angle chasing and a little trigonometry one could conjure up a function that spits out the total length of this contraption based on the input angle.
Then it's only a matter of checking if it's constant or not
We are all about to witness another historical event.
these videos are so unassuming but so interesting
Man, extendo boxing gloves are gonna be all kinds of newfangled fun.
All these comments talking about computers and technical stuff, then there's me not understanding any of that and thinking "This looks like a fun thing to fidget with"
It's so silly, why do I get excited from such an impractical way of constructing logic gates haha
You manage to blow my mind, everytime!! 🔥❤️
Incredible. My brain is satisfied with tonight's TH-cam dive before slumber.
After this video I'm sure it won't be useless anymore. This was really cool.
I feel like if you wanted more mechanically interesting movements then I would add additional layers. So the same scissoring motion would produce different movements on different layers. I feel like that would lead to more possibility for combos and computations. Could possibly make for a fun calculator where you flip some switches and then pull a big lever raising the scissors up to compute the answer.
This is fascinating! I'd love to see a basic computer system built out of these linkages.
I was already interested when I saw the thumbnail but I saw that L joint in the first 5 seconds. that's awesome
The harder question is, can you make every gate, and just a wire, with constant length, or else your entire system will break when something changes length…
With two constant length NOT gates you can make a constant length wire.
Could you please guys check my comment if the AND with three NOT gates could work? thanks
Thanks for sharing this, it takes me back to 9th grade science class. Very Kool
I'm thinking if we do make it, we'd probably need a huge space to do it first, like how early computers were vs how they are now (entire basketball courts vs your typical smart phone).
I think one can create an XOR gate by using 4 of those NOT scissors.
First, place 2 of those NOT scissors parallel to each other.
Then, connect 1 NOT scissors perpendicular to them at one end. Now it would look like a コ shape. This way, it would shift shapes left or right only when one input is True, but jams at the middle when both are True.
Finally, attach a short stick to that sideways NOT scissors' middle point, and place it in between the 4th NOT scissors' arms. It will open when the perpendicular one shift shapes. Attach a small circle at the pivot point of the 4th NOT scissors' arms. This makes sure the stick pushes the 4th NOT scissors back in place once it's opened.
Would successfully setting one input not simply modify the other input to be the opposite?
@@Owen-bk5fc I haven't thought about what would happen when the inputs aren't changed simultaneously.
That might happen if both inputs' arms were directly connected to the perpendicular one. I'm still thinking about a solution to isolate them without over complicating it with strings and rubber bands and stuff.
@@Owen-bk5fc Currently, the (00) and (11) state works by balancing the opposing forces from the two inputs.
So if you pinch one of the inputs halfway through, it can change into that (00)/(11) state you expect; but if you pinch too hard then it modifies the other input too.
Very interesting. Thanks for sharing this.
To make it constant length I would assume you could just increase the thickness of the parts to they don’t close at such a tightly angle
imagine a robot using these as muscles
"I've never seen this used before, possibly because it's kind of useless" amazing
This dude is exactly my kinda guy
Going on pure instinct, for the constant length; two ideas figuring them out as I type.
The movement comes from the rotation of the connection line between the axes: it's a square pivoting on a corner with the axes on the connected corners
You need to get the overall rotation point in between the axes. But with a fixed axis that won't work.
So: have that move along a curved path, sliding along a shaped "linear" axis.
Or
Extending the length of the first/last joints changes how agressively it extends. It feels like that should help somehow.
There's lots of "interesting mechanical mechanisms" demonstrated on here, many dedicated to hinging arms (like on the channel tenfold). there's probably one that is applicable here?
What is also interesting is that you also have a kind of bit here: left or right, somewhat fixed. That's a 0/1
I'd like to make note that a one link scissor is a Not gate just shifted by 45 degrees. This is because as one opens the scissor it closes the sides. This should help simplify any scissor circuit.
its interesting people find these logic mechanisms everywhere, why nobody ever built anything out of them is reliability, even in the days of analogue computers, they tried everything but cams, gears and rods were the most reliable versions that could run indefinitely
Hey Henry
- long time, no see.
Not done the analysis, but taking a lead from electrical theory, three phase might work for lengths - i.e. two sections of 120 degrees scissors, with a longer middle section. Trouble is, that’d destroy the not and make it a “buffer” again - like the top one - but one of constant length.
I’ve been idly mulling the “and” issue - nothing truly elegant is presenting itself. I hope that gets solved - it’d be nice to see even a full adder. I suspect friction will start to be an issue!
Just figured out how to make simple OR/AND gates :
Arrange 2 NOT scissors and 1 regular scissors into an 工 shape. Both NOT scissors' middle points connect to one end of the regular scissors.
Add a third NOT scissors and form a 王 shape, its middle point connecting to the regular scissors' one.
Now depending on which direction you shift the third one's position, you can make either an OR gate or an AND gate.
Notice when one of the inputs is True, it stretches the regular scissors and forms a Z shape. By moving the third NOT scissors slightly to the left or right, you're essentially flooring or ceiling that 0.5 state into a 0 or 1.
That's epic! I don't think that would be useless. That would make a very cool entry/security gate for a residence or storefront.
A pair of bolt cutters is basically two of these in series, without the excess scissor bits on either end.
Fun thing about a not gate. If you put it's output back into it's input it oscillates. I imagine because this mechanism has a neutral state it would just be stuck at neutral.
Great video!
You could make analog, or fractional inputs, out of the relative lengths of parts, this could be very useful
Just found this video. Liked it enough to subscribe after just one video. Very interesting idea. I would love to see a scissor link computer. (Also had to like and comment for the TH-cam engagement)
Some kind of clutch would be a useful building block for an AND gate or a CNOT gate. One input acts as a control to allow or prevent propagation of signals in one or more target linkages.
I could imagine this working by coupling the target input and output by way of a broken pair of straight linkages (not scissored). The control input is mounted orthogonally to the target and acts to either hold the broken linkages straight and parallel to one another (allowing the signal to pass) or by letting them buckle in the middle (preventing transfer of the signal). A single control input could link control multiple targets simultaneously which would be useful for multiplexing.
To make an AND out of a clutch you can either add a spring to provide some restoring force in the target output when the clutch is disengaged or combine a complementary pair of controlled gates in some other way.
Dunno why I clicked, but it was worth my time
I cant imagine how a computer made of scissors would be
Cool
Thats genious! A 90 angle is kind of like a not gate to a straight line
That 90 degree piece has a name, it's a bell crank. I highly recommend you check out @ArglinV3 and @thang010146 for linkages that perform logical operations.
If you want to create a "pass" node, that outputs its input value (basically a wire) in a similar manner, you should connect two (or any even number) NOTs, and assuming the constant length problem is solved, these wires would also be constant length. Also they would be really cool visually, blending in in a larger machine. This would also have greater mechanical strength than some of the other ideas I have for just wires, and these can actually be suspended in any direction.
Another huge problem with the future designs is buffering. You can manage to create logical blocks that perform a few operations as gates, but after a certain complexity you either need very good bearings, which is an expensive option (and still not infinitely scalable), or some kind of logical buffers, that "amplify the signal" passed to it. This is a pretty straightforward electromechanical challenge, but the whole point is to only use this reduced set of building blocks. I'm thinking maybe something with "rechargeable" torsion springs and latches like in a mechanical clock.
Brilliant!
That's crazy. You're onto something
Fascinating
Mechanical logic is fascinating
'These two look like they're identical mechanisms'
Me: *Proceeds to pause the video and find the difference between the two* (and found the difference)
That is awesome.
some notes:
- i think the grooves idea like below is a nice way to make the length constant
- it seems like you'd more or less be limited to trigonometry because of how scissor linkages work. and you can't really make them add up to a constant. there's the good old pythagorean theorem but that requires sin^2 and cos^2 so that doesn't work either
- you *may* be able to get away with having some curved linkages - potentially just at the ends, potentially differently curved for each half (plus curved L pieces - if each half is differently curved you may (or may not) want the L pieces to be asymmetric)
- for 2-input gates you'd probably want some kind of junction. either coming in from three sides of a square, or three sides of a triangle
- for convenience (well, to be able to make complex circuits) you would likely also want a simple mechanism to rotate a truth value by say, 90 or 120 degrees
- (i'm not sure how logic gates would work - you may need additional degrees of freedom like the grooves?)
Very nice!
For and gate, two not gates beside each other with the inner outputs linked with the identity linkage? (Outer ends of the not gates fixed I suppose)
It wouldn’t be perfect in terms of deciding power - both off would be markedly different from one on would be markedly different from two on - but it should be closer maybe?
Thanks for sharing Henry. Hope you can follow up on that newspaper article and expand on that world mechanism. I'm very curious.
I'm not a math guy, but I think you can also create a signal inverter by introducing an orthogonal junction in the links - as you expand one dimension, you contract the other. You might be able to create a parallel inverter with the use of this pseudo-NOT gate's constant length. You could create an absolute NOT gate by eliminating the second half of the links to terminate the extension throughput.
I would reframe the whole mechanic a bit: suppose one arm of the input is fixed to a grid, then you only look at the movement of the other arm. This reduces the output to one vertical movement too. Now you could imagine a rectangular frame that can only move vertically. The horizontal movement of the scissors along the horizontal part of the rectangle is ignored, you just have some leeway for the length change there. The vertical movement is the output you are interested, you get that from only one point of the two output scissors, the other one is moving in a horizontal rail to give it leeway for the length change, but it is fixed in the vertical axis. Those frames are the "wires" that you use to connect to the next stage of logic gates.
It's not a fully scissor link mechanism this way, but building a first prototype with those moving frames could give you some insight in how a logic circuit could work. Maybe you find a scissor variation of that idea later.
If we're aiming to not do anything too cheeky (like a corrector segment at the end), it seems to me like your inconsistent length is caused by 6x the hypotenuse of a right angle triangle made by the line segments being longer than 6x the length of the line segments. The obvious conceptual answer to this is to somehow change things so that we're measuring equilateral triangles instead, but I can't for the life of me picture what that would look like.
this is GENIUS
Bro just acualy blew my mind
if you can measure the distance between the arms on each side, and hook it up to an osilating mechanism, you can make an encoder for counting whatever causes the osilations.
You could extract two bits of information ( ie. Light which goes through the scissors and the length variable for each scissor. Although the light would be position dependent for the second scissor.
you know .... if you put a Huge Red Everlast boxing glove on one end and a trigger mechanism on the other -- you could get quite a laugh going .
I'm scared of what this may lead to, and I love it
Just an idea for an and gate but it looks like you could make a linkage off of the middle 2 ends to get what seems to be kind of like an and gate
You are now legally obligated to make a computer out of them now
All I can think is how that would be great for those hard to reach circular clips that standard pliers can't quite reach.
People really be using everything to make logic gates, except of transistors.
I’ve seen it with water, fireworks, RC cars, many other mechanical constructions and now even with scissors. 😂
Can't wait for the xor gate
looking forward to the nand gate
will be waiting for the half adder video
I wonder how stable it could be at the center L joint. Since it’s anchored at both ends, and can be driven from something stationary. Could be an interesting choice for something like an extrusion 3D printer.
Or maybe since it converts a very long motion into a comparatively fine one it could be a tool for firmly but precisely gripping fine objects.
You say scissors based mechanical computer like it's some kind of joke, but there is reason to believe that the cytoskeleton(a cross linkage of various proteins that support the form, movement, and metabolism of any cell), besides simply acting as scaffolding, is actually encoding information and performing computations which direct the cell's metabolism and behavior. This can explain how a single-celled organism can have fairly complex behavior, but also how cells organize themselves in general. I bet that the some of the proteins that make up the cytoskeleton can have distinct configuration states, some which are just like the scissor arm which can be elongated or compressed, and there is a some kind of junction protein which acts similarly to your not gate.
In order to make one of (Very close to) constant length you would need a link that is at the middle point when the input is open and open when the input is closed.
I saw the title thinking "yes those are indeed scissors and not gates"
I was immediately thinking that going through 3rd dimension is a quick lazy answer. Add hinges to the arms so ends of the mechanism stay a fixed (arbitrarily adjustable) distance from each other but the middle of it bows a up and down, freely, in sort of an arch or wrinkled shape.
Someone is going to make bad apple in a scissor computer
Please send this to Steve Mould, he would absolutely make a computer out of these for eg. simple mathematical operations.
instead of an and gate, you could also make an or gate, since nor gates can be used to make the other logic gates, like a nand gate
I like that this allows you to output true to true and false to false without changing the overall length (too much) as compared to a "normal" scissor linkage. Not sure of a practical use, but it is interesting :)
It is now only a matter of time before we see the scissor linkage 4 bit adder.
Just like that though you have segmented/concertina locomotion in a non-biological way. Super cool!
If you had a not gate of constant width, you could store bits in constant length "wiring" by just having not-not.
You can make every classic gate by combining a bunch of not and or gates. I imagine making an or gate might be easier
i remember you could do logic with just or, not and falsum. Or rhe duals of course.
I have looked at a TH-cam search and there seems be a LEGO Technic implementation
If you manage to design either a NOR gate or NAND gate, you could construct any gate from those because they each are functionally complete
Well actually it's either NAND, NOR, AND or OR, since you can use NOT gates to convert between those 4.
Could you please guys check my comment if the AND with three NOT gates could work? thanks
It reminds me of the giant robot tarantula thing from wild wild west,
Like it would be used to make opposing legs move at opposite times or something
"I've never seen this used before!"
"... Probably because it's kinda useless"
just hilarious 😂
Nice!
sick video
have you tried leftangles for the linkage? 60°?
This the type of stupid I like. Now I want the other gates as well.