Testing if Marbles can play Tight Music

[@poudreuse8391]

The enemy of good is better. You need to predefine the tolerances you want. It appears you might already may have achieved them as it better than you are currently.

[@Wintergatan]

Thanks for the comment, i see this sentiment a lot and i 100% agree in principle. I need to watch out for the point of diminishing returns. But,Here are some reasons why i want to keep on researching tightness and see If i can improve the results from where they are today: 1. I have barely started figuring out What the Huygen drive is capable of, i think There are low hanging fruit that i do not want to miss. 2. Original MM and MMX failed because they had inherent design flaws Built in from the very beginning. I do not want to make this same mistake again. The tightness is a non - negotiable design requirement. 3. Predefining the tolerances, yes but not until i am somewhat familiar with What the situation is. A random number currently would not help the process. But yes, i really need to accept some error. If i am going for 0.00001 ms this is impossible. I think i Will settle for around sub 1 ms But currently, my process wouldnt look different If i chose a random tolerance. If i chose 0,5 or 2 or 1 the design process would be identical. My drive comes from the Exciting process of seeing How accurate i can design this, and i Will need to shift from that Mindset to a ”great is enough” mindset eventually. 4. For the Two previous machines i almost always had to quantize the audio in the computer afterwards. If i am going to want to make Music with the new Machine, it needs to play tight. When i say that i Will cancel the project of it isnt playing tight, i am serious :) I wanted to take the chance To share my thoughts on this as ypur comment reflects the thoughts of many viewers. The Machine is already performing very promising. But Prototyping a bit deeper to fully understand everything is a must for the success of the project👍 EDIT: 5. A human musical timing has feeling, "errors" is often expression and groove. The errors from a machine are just errors, they do not express feeling or groove or finer aspects of musical timing like that. If i can make the marble machine play really tight, i can program that feeling in by deliberately offsetting marbles from the perfect timing. Perhaps some snare hit should be early or late to give that special groove for that exact beat. But in order to do that the machine needs to be able to play when i want it to. So, just playing better than me doesnt really cut it as the machine errors are much worse for the music than the human errors.

[@jhonny44444]

@@Wintergatan In all design projects you need to start with requirements. You already came up with a lot of them but in respect to your timings/tolerances, having a requirement that provides a MVP target will give you a baseline from which you can build. I 100% agree with you that proper research, be it based on literature or based on experimentation, is key as you might miss possible solutions that could've been the best if you jump onto the first one that works. But as all projects have a budget and a time limit, you need to know when to freeze design decisions and move on to implementation. Looking at your experiment in this video, you already described your MVP, i.e. yourself. You are capable of producing beautiful music with your own tempo capabilities, so, a machine that can match your skills can do the same. Seeing as your design is already better than you, optimizing the design might be all that is needed:) (more inertia (you will need it if you have a lot more - variable - load on the system), less friction, less bending due to better mechanical tolerances etc.) I've been following your project since the very first video you posted about the primary MM and loved watching you develop the machine and yourself in the process, I'd hate to see you tumble down the perfection rabbit hole! Cheers and I'm, as always, looking forward to your next video!

[@XeroShifter]

I think you should work on trying to find out how tight an average person's perception of a beat is. Its pointless to try to optimize far beyond typical human perception because it will make no difference to someone's experience. You of course shoot for better than that number in the end for the sake of those with extremely good perception, and to ensure that situations that happen outside of a single standard deviation are still not terribly disruptive, but if you had the human benchmark you'd have a better idea of where "great" really is. If you need a place to start, another music youtuber Adam Neely might be able to point you in the right direction as he has touched on human timing perception before. There is fun to be had optimizing and engineering; and exploring the many options out there is wonderful, and I look forward to that journey with you. It would just be really sad to set the sights on some arbitrary number like 1.5ms SD, and fail that, but not continue the project over it, while the limit of human perception could be something like 5ms. @@Wintergatan

[@hadrast]

For reference, I have seen ±3㎳ used as a tolerance benchmark for audio syncing in various contexts, and have verified anecdotally that that is in fact a good number to use for syncing audio/visuals to music (for sane tempos, at least).

[@ChrisLindsley]

@@jhonny44444 You're mostly right. You need specs before yuo design a project, but where do your sepcs come from? You need to start with research (market, scientific, sociological, and prototyping) to find your specs. I think that's what he's doing now. However, I do agree that some more research is needed outside of prototyping. Perhaps having a friend varying a click track randomly by increasingly larger standard deviation until Martin have detect it.

[@Zeragamba]

4:36 It would have been *really* helpful for us to hear what the MMX sounded like unedited so that we can better understand why you felt this way

[@namenull7399]

He might not have the raw audio anymore.

[@Sthunderrocker]

Would have been a lot more helpful if he hadn't misrepresented it all the way along

[@darkforce2803]

I don't think Martin misrepresented the MMX all that much, given that it also worked during live streams without post processing. Sounds more like a coping/defense strategy tbh.

[@gastonbell108]

"But Professore DaVinci, your steam-powered flying machine was almost finished!" "Flawed and disappointing. To the trash heap it goes."

[@TheFlyingSailorYT]

@@Sthunderrocker What are you implying?

[@nutnurut4260]

Just a handy tip: the minimum number of teeth you want in a gear is 17 due to durability reasons. Also having a low common divisor between the gear teeth will lead to some teeth being more worn since they interact with the same teeth and at the same force on repeat, try to have teeth ratios like 17 to 37. Should lead to less need for maintenance down the line

[@EaglePicking]

The numerophile in me loves this :-)

[@hyperteleXii]

So prime number ratios?

[@EaglePicking]

@@hyperteleXii Primes have the lowest possible common divisor indeed.

[@WorBlux]

@@hyperteleXii Co-prime number ratios actually. You want numbers with no/few common divisors. - (8,5) (35, 36), (10, 21) all work. By definition every prime is co-prime with every other so any prime number of teeth work, but makes it hard if you are aiming for specific ratios. If you want say 8:1 160:21 or 81:10 are pretty close and wear a lot more evenly - every tooth on the small gear will engage every tooth on the large gear every 10 or 21 rotations. Wheras with 80:10 or 160:20 every small tooth is engaging the same 8 teeth on every rotation. Any flaw in the gear meshing then gets amplified.

[@southernflatland]

But he's also looking for the sound of the gears to fall in time with the music. I don't think that'll work out with prime number gearing..

[@Vanguard1262]

I’m glad you actually set a human benchmark to set the expectation, instead of setting such an unnecessary one of

[@TheJoukai]

Yeah, it really bothered me in the past when he was like "10ms off? Unlistenable"; glad to see this test is grounded more in reality.

[@YogiTheBearMan]

I think part of it is 10ms per part of the machine is going to add up to more than 10ms overall, as well as easier to track down where in the machine the problem is if each part is tested very tight

[@ouzoloves]

Especially as such a large part of music is the feeling that the human element gives.

[@catgirlQueer]

for individual components getting the variance off will add up, a ±10ms and another ±10ms is quickly ±20ms

[@Steeger007]

Music shouldn't be measured by a computer, it's measured by human ear and human element. Remember after he adds the marble machine and all functioning parts, he's still playing with other people.

[@TofuRabbit]

I'm not gonna lie, every time Martin shows the MMX all i can think about is how much i miss it. The smaller footprint/more compact design, the metal and wood and transparent pipes, it all looked amazing and beautiful. I just really really wish it worked out.

[@beansnrice321]

You might like the technique Steve Mould showed off on how to enhance video motion in a way to reveal subtle vibrations the human eye has a difficult time seeing. That last shot you showed of the gears winding and flexing made me think that you could probably literally see where the efficiency is being lost with the technique Steve was showing.

[@Scanlaid]

Not sure Martin could afford an enterprise-level camera like that... but he should contact Steve and see if he can get a relationship with RDI for a loaner camera. They might love a little advertising to see how their technology could apply in an instrument manufacturing context? I'd love to see one of those cameras film any plucked or struck instrument.

[@nickfosterxx]

Great thinking.

[@Fractal2D]

"Eulerian Video Magnification for Revealing Subtle Changes in the World" is the open-source version of that.

[@dfgdfg_]

Steve Mold collab +1. He's in the UK

[@dranorter]

Since the beginning of "video magnifying" tech, it's been available for free on the web. I'm glad people are making fancy hardware for it but it's only needed if you want to detect vibrations of higher frequencies which get hidden by a normal frame rate.@@Scanlaid

[@blacklistnr1]

digital artists: add some randomness to make sound more human mechanical artist: tighter!

[@InventorZahran]

Pat Metheny (electro-mechanical artist): digital precision on a mechanical orchestrion controlled by a human

[@danielboyle435]

@@InventorZahran Pat Metheny is an absolute legend. Incredible guitar player and his innovations to music are spectacular

[@victorcercasin]

Damn, I think Martin's idea of showing the MMX machine video was to show us how bad it was, but it really just showed how awesome it looked. I really wish it had worked out.

[@doro5751]

If you check out the @musikkabinett channel, you can actually see a working MMX

[@EaglePicking]

@@doro5751 Wonderful how the community there made the MMX what it is supposed to be: a museum piece. (and not a music instrument to play live on stage)

[@doro5751]

@@EaglePicking yes, we are trying our best.

[@darksmile4719]

looked cool but it was loud and had so many issues

[@BenjaminJBromley]

I rarely him talk about weight, ease of setup, or portability as design requirements. Makes me think he really doesn’t intend to take it on tour. Just get a raspberry pi, some relays and servos and get on the road already, Martin!

[@GabeDelaSoul]

This is starting to feel like the chase of perfection all over again. "If I cannot play sub X milisecond music, then I will call it quits." Like an unhealthy fixation which stunts how worthwhile this project truly is. Chasing an idealized version of perfection, without calling it such. I think a ton of us considered you a winner already being under the 5ms range, it's bonkers. Degrading your previous works as duct taped hunks of wood just devalues every single person's opinion and support which got you to where you are right now. No one cared that the original had imperfections, as no one can count how many miliseconds were off on one impact to the other, because the culmination is greater than the sum of its beats

[@renierelagazo7888]

Same thoughts.

[@randomnickify]

He is musician, what sounds fine for you sounds like s**t for him. Painter will not work with pencil that can not paint straight, Driver will not race in a car that cant corner properly.

[@corneliuslow]

​@@randomnickifyexcept music is not necessarily better when the notes are perfectly quantized/“on-time” with a metronome. Less-than-perfect timing makes it more lively and human, and is often THE key to good feel and groove.

[@zusurs]

@@randomnickify Bulls**t. You sound like one of those fanatics who says that "gaming monitor must be at least 240fps, otherwise it's UNPLAYABLE". Just STOP already. No musician on the damn planet can actually hear a 5 millisecond deviation, let alone a sum-one ms deviation. Just stop defending that bullsh*t.

[@BenjaminJBromley]

I’m a musician and am often intentionally trying make computer music sound less perfect. Martin phase 3 makes me slap my forehead. I can’t even get through these videos anymore, as much as I want to see him succeed.

The test at the end, when you checked that it sounds good, was in my humble opinion the most important thing. And it feels tight.

[@scaredyfish]

Amen

[@strikeforce4755]

Hello Martin. Just wanted to say as a drummer and a percussionist i appreciate the talk about tight music. For me tight music means playing close enough to the tempo so that it is not noticable off beat or slowing down or speeding up. With all of this discussion of tight music i think it would be good for to to come up with/state what you exactly mean by tight music and then maybe try to determine what is the threshold for good enough tight music. Perfection is great but sometimes good enough is all you need cheers

[@DanLocks]

He mentioned when generating his data that "a professional could do better." I'd love to get some data from a pro and use that as a the specification for tight. Additionally, how much control does a pro have for reliably pushing ahead or behind the beat? This data would provide goals for the musicality he's looking for.

[@nick_cnc]

best part was Martin admitting that the audio itself sounds "very, very, good, actually". hopefully he's starting to realize that perfect numbers on a spreadsheet aren't the most important aspect of building a music machine.

[@MicahBBurke]

Yessssss

[@DJRockinRob]

I would like to see some "Standard Deviation" testing on some of the Music Machines from the museum... They were successful PRODUCTION machines... I bet they dont play as tight as he is trying to play with the MM ;-)

[@mrscottcheek]

Or how about tests of other professional music. I feel like I hear songs on the radio all the time with sub perfect timing.

[@Knowbody42]

It's strange, there have been design choices he made that other people criticized for not being ideal, and Martin said that it was good enough. Now it kind of seems like he's starting to allow perfect to be the enemy of good.

[@LetUsPlayMoreGames]

I was commissioned years ago to put someones sheet music into a computer to get a Master recording. They hated it because it was too perfect and sounded robotic. They said it didn't have feeling like when a human played it. So perfect sound does not mean perfect music. It reminds me of the argument between Analog and Digital.

[@RegebroRepairs]

I'm doing tests with MIDI timing recently, and all your tests are tighter than a computer with Ableton live sending MIDI data. 😀

[@bitethebeat_music]

Oh! I'm currently exploring and learning sending MIDI from Ableton to Resolume Wire to sync visuals. Fun!

[@thomasbecker9676]

So you're saying that the numbers Martin has been chasing for what, 2 years, are unachievable?

[@jcagle31337]

the real underlying thing that is helpful here is that all the freaking 'can this play tight music' is really just a defense mechanism to not having to actually build a damn marble machine.

[@youareliedtobythemedia]

Use the windows midi streaming api

[@Nevir202]

Ableton seems to be quite a laggy program TBH. Just running a few audio filtering and FX plugins on recorded audio, I've seen as much as 500ms latency

[@SuperAWaC]

watching the ever deeper descent into madness and obsession with perfection. i wonder when the last time i actually heard music on this channel was, and if i'll ever hear it here again?

[@chrisfedde4032]

You still need an active speed governor. Something that actively adjust the speed based on the system load and available power. Like a CVT.

[@eloc580]

cvts and other governors still change speed with load, just not as much as an ungoverned system. a computer control would get the best speed control (like a small brake or motor to add/remove rpm) a cvt would be better than a direct drive though

[@John_Weiss]

I've been thinking exactly this for months now! He needs a CVT that couples the power-module to the rest of the machine. You can't expect your power-module to provide tempo. My phone's Li-Polymer battery _cannot_ keep the phone's clock synchronized. There's dedicated circuitry and software for keeping the clock in sync with a central time source. The battery's there to provide power, and nothing more.

[@EaglePicking]

I'm sure Martin understands what a governor is by now. He's got a simple air governor on the prototype already and I'm assuming that he is already researching and contemplating what kind over governor to implement in the final design, because he wants that thing TIGHT ;-) The ultimate governor for the MM3 would be one where the tempo is kept by simple physics, not electronics, and where it could be set with some sort of steampunk-looking lever. Some sort of adjustable CVT would be great.

[@Oeli_Maurer]

Exactly! If a single marble channel‘s resistance can already throw off the tempo, dozens of them will cause the tempo to be all over the place. The problem is, that constant force is applied to the gear train. If there‘s more load, it will obviously turn slower. A centrifugal govenor will solve this problem because it supplies variable force depending on the load to achieve a constant tempo. Also, a heavy flywheel dampens the action of the govenor, so he should eighter get rid of it completely or just use a very light one.

[@Alchemistake]

Cvt has too much loss

[@Norrin277]

All I’ll say is this. There is perfection in imperfection. Don’t loose your spark by trying to eliminate every single tiny imperfection.

[@newolde1]

I also think that much of the modern money based society has frightfully downplayed the fact that play is an essential aspect of our lives. It drives our creativity, the ability to reframe our perspectives, and refine our core values.

[@Dysonsphere42]

FIRST FLOOR MARBLE!! id say things are going good

[@simonstergaard]

next machine should be named MarblemachineM ...as in roman numeral M...not X.

[@fmga]

@@simonstergaardthen it could be mmm or m3 (works on two levels bc m^3 is the same as MMM mathematically but also bc the m3 would be the 3rd marble machine)

[@RyleaLynn706]

With any luck, it'll be the *only* floor marble of 2023.

[@glenmcgillivray4707]

@@RyleaLynn706 no no. We want floor marbles during the testing phase as it shows us the failure points before we finish building up. Better now than later.

[@RyleaLynn706]

@@glenmcgillivray4707 To be fair, 2023 has like a month and a half left before it ends. There'll be plenty of room in 2024 for floor marbles.

[@toddcamnitz6164]

Start of the video: “let’s see how tight a human can play.” The long-time audience: “oh thank god. At last.”

[@rudie2902]

Martin, I think part of your varying results is that you use the bike brake cables for pushing. They are meant for pulling only. To achieve this you need to put a spring at each receiving end that you can pull back with the cable.

[@IanZainea1990]

9:13 mock up. It is a mock up. A prototype is a finished item. Like a car prototype, it's the full vehicle but made by hand or manual processes, used to make the final product in mass production

[@KinderAndry]

When you tested the marble drop, you noticed that angle of plywood, on which marbles are dropping, is affecting the standard deviation. The angle has to be 0, otherwise marbles could drop on slightly different positions and adding the error.

[@chevrotain]

but if you think about it, the marbles will always be dropped on instruments with an angle, in order to control their fall. as long as the angle is consistent then the tests can be conclusive. the angle should be considered heavily though, you're right. perhaps some testing in the future on the perfect angle for instruments ... ?

[@PapaLurts]

but if you have an angle of 0 marbles will not bounce off the board and give additional readings

[@DaShikuXI]

@@chevrotain Not just the angle but also the fact that the instruments will be vibrating, which will constantly change the angle.

[@chevrotain]

​@@DaShikuXItrue! lots of variables to consider.....

[@glenmcgillivray4707]

​@@DaShikuXIthere's a reason I've made a suggestion that droppers and instruments should be on a seperate frame isolated from lifts, fear trains, and even the programming wheels, to minimise vibration or variation of positioning. Maximum consistency. Minimum external disruption. If the funnels or marble lifts are vibrating a little, it makes no difference to marble drop position, timing, mechanical function, or operation of the machine. While vibration at the mission critical instrument or dropper elements will create variation in music and Martin losing hope once again. Welding the busy flexing and vibrating marble and music machine to the elements that need to sit still to make music is not going to fix past mistakes. Ideally they should be installed such that when lifted the sub frames lift with the core construction for ease of use, maybe positioned with V notches to settle down into. Yet be freestanding when placed down. So the only mechanical connection to the marble machine the instrument components ever deal with is floor vibrations. Which would be beyond the scope of this project to ever isolate from. What would you do? Add bathtubs to let them float, and rubber bands to maintain consistent positioning? Let's not go off the deep end. Although it would be funny to watch the pirated instruments Marble Machine design.

[@fumitakahayashi]

It's like Martin is looking for an excuse to quit, and he can't find it.

[@Nooticus]

THIS.

[@ShannonCarey0x53]

Or an excuse for continuing to publish KZbin videos forever.

[@BlueCubeSociety]

and he's always looking for an excuse to bash on MMX, even though it wasn't as bad as he thinks it is, and I would guess most viewers liked the old machine regardless how tight it is

[@peterdilworth3110]

If you ever bothered to invent something you would find his journey all to familiar

[@AdariousMistdancer]

Too much second guessing and too much seeking of perfection, can and will result in a project that never gets completed.

[@SKIDOUG58]

I think we are at a point where as long as the marbles don't get hung up, end up on the floor , and are doing what they need to do we are good to go ...perfection is not what music is about...it's soul

[@c0mpu73rguy]

Ehm, again, the interpretation doesn’t need to be perfect but the instrument, especially if mechanical like that better be (unless the builder himself is playing said instrument and knows all its quircks like Martin).

[@slawinsky8951]

Your opponent is not the Marbele Machine, but your obsessive search for perfection. With the first machine, you still managed to outwit this compulsion. Will you be able to do it again?

[@Endoe.McKronic]

NO! He'll just keep asking people for money. And all those parts that people sent to him for the last machine and he just trashed it and let some other group of German's finish it for him. Now its in some museum of failed music!!

[@TheBlackAbyss1]

​@@Endoe.McKronichow is he asking for money? He cancelled his patreon out of respect and guilt of people continuing to pay for a failed project

[@d4slaimless]

@@TheBlackAbyss1 Well, in his last video before the break he said he now have enough to finish the machine. But instead he went for a vacation, changed location and sent the machine to some other people. Good thing is that guy knows when to stop. Bad thing - he didn't follow his own words. And he only let everyone know already after it happened. Hey guys, I decided not to do it. Hmmm..... More good things about Martin's projects - he keeps entertaining people with this videos. And this is what youtube is for. One more really bad thing - in this exactly video he admits that MMX wasn't actually playing much, he was just making it look cool. But then what exactly were people following and donating their money for? They were mislead to think the project is going somewhere, but in actuality it was an exercise in futility. Under some of the early videos someone posted a prediction that the machine is never going to be finished. And maybe the process itself is more interesting, but project that never ends and keeps recycling itself, while also not being completely honest in the process... But don't get me wrong. Martin is a content creator and if people like watching those videos and have their own creativity stimulated, then there is something good in it.

[@davidanderson5310]

I think the reason the extra weight makes the tempo deviation worse is that the fan governor works best at a particular speed. You'd need to resize it for different speeds.

[@abelnemeth4346]

Also, the flywheel might have been too uneven.

[@AdrienGagnonSoundTech]

@@abelnemeth4346 It's for sure an issue with the flywheel. If you have even just the slightest weight imbalance (caused by the odd shaped metal bands and added screws) then the weight distribution of the flywheel becomes uneven and that's why he had tempo deviations. The flywheel is now making micro wobbles due to the added weight, and it throws the tempo off. If you want added weight, then you need make the wheel out of a heavier uniform material, or make a larger wheel.

[@nashpark7719]

Like a lot of people here it seems like you're chasing perfection in something that does not need to be perfect. If this was a piece of medical or aerospace technology it would be worthwhile - but as an instrument I'm not so sure. People would've been thrilled to see you playing the MMX on a world tour with marbles flying all over the floor!

[@voidtv8401]

I think you’re finally starting to feel something you’ve needed to realize for a long time. You’ve been trying to make a science of it all, but in the end, it’s music. It’s not about what’s perfect: it’s about how the sound makes you feel.

[@c0mpu73rguy]

Playing an instrument is music, but building one is still mostly science (and technique).

[@sbea4002]

@@c0mpu73rguy I disagree, if we aren't able to make an instrument we use our voices. If we don't have the ability to speak, then we tap. If we can't tap, then we move our bodies to make a rhythm. Music is in the soul and the more you try to automate it, the further away from the soul you get

[@c0mpu73rguy]

@@sbea4002 Tell that to a saxophone player and see his reaction. It's not because something is complex to build that it doesn't make music. Yet I'm pretty sure most saxophone players couldn't build one and that's okay.

[@tonymahboi]

Shout-out to the discontinued IKEA Algot gang!

[@Wintergatan]

I will never get over it.

[@albex8717]

The fact that the test results got worse and worse for a while there might imply that there's some component of the prototype that's wearing out fast and thus skewing the results

[@ivandagr857]

I was thinking the medal bands distributed weight uneven

[@t.m.w.a.s.6809]

I want to say this for the viewers getting upset about his dedication to tight music here. Firstly, Martin made a LOT of mistakes early on, something that he readily admitted many times, and that made the early part of this big long marble machine project where all the hype was at a peak kinda doomed to begin falling off before it would be finished. Now, a lot of people are saying that they are disappointed with his obsession with tight music, thinking that the audience will lose interest by the time the marble machine is done, etc. but Martin has been learning the whole way and has done a REALLY good job of refining his engineering skills. This whole project, to make a marble machine that can go on tour, hinges on a few things: 1) Ruggedness of the machine to survive a big long trip around multiple states. 2) Workability for repair for the eventual damages it will go through on said tour. 3) A very tight and precise core of the programming wheel to feed into the other instruments. Why does it need to be so tight? Not particularly because it will sound HORRENDOUS if its only a few milliseconds off on a note, but because of the programming wheel being the core, the center of where all the other instruments get fed their timing, so if the programming wheel isn’t tight on the timing, then the inevitable imperfections of the rest of the mechanical structure that feeds into each instrument will be enough to make that slight inaccuracy grow into a very noticeable, off time, and inconsistent tempo. I think Martin has it down here, only a few kinks to work out in terms of why greater inertia isn’t performing better, what length of the programing slots is necessary for different rotation speeds, etc., but guys, do you see what he’s making? Do you see how far he’s come to engineer all this effectively from scratch? Im proud of him. I agree that a lot of the hype is definitely quieting down, but i still love seeing his videos in my feed and really want to see that work of art on the stage. Martin, if your reading this, you’ve got this man, and even if life happens and you cant complete this, you’ve done way more than I’d initially expected when i first saw your plans to go on tour. I love ya.

[@toydotgame]

This is a very very tense point in time to be involved in, seeing the fate of this project I've followed for almost 4 years now be decided. I really want to see a MM3 come to fruition; there's a lot of obsession with perfect reliability and tightness in numbers, but in the end if a machine is created that can play more than one song in a row, it'd've beaten the MM1 and truly succeeded in my eyes.

[@Zichqec]

If it can play a single song the whole way through, it will already have beaten MM1

[@justinnaramor6050]

And you, just like millions of other folks, forget that the first MM couldn't even play a single song "all the way through", reliably. Lol! It's so amusing how people are so misled by that first MM video. Like another commenter said in reply to you, if this new machine can play even one song "all the way through", reliably, it's already beaten the first MM by a long, long shot.

[@toydotgame]

@@justinnaramor6050 I am aware of that fact??? If the MM3 can play a whole song or more then it's succeeded in my eyes, that's all I was saying.

[@justinnaramor6050]

@@toydotgame OK, good. My apologies if I misunderstood. Still, many people still seem to think that first marble machine played only the one song reliably and then failed later. No, it never worked reliably at all, even while attempting to play that first song. Someone's gotta clear up this misconception sooner or later :).

[@FJX2000_Productions]

I always thought the goal from the beginning was to just make an impressive marble machine that could make music. If that was the goal, then both the original Marble Machine and the MMX were successes. It pains me every time Martin calls them “failures” just because of how much the fans loved both of them, and it’s almost a slap in the face really with just how much effort and collaborative man-hours/resources went into the MMX. But at some point I think Martin decided these machines needed to be “perfect” or else he was failing, and to us as fans, that was never even close to our expectations. We just wanted a cool machine that could make music. There is no such thing as a perfect music machine, there is no perfect creation. It’s like designing a perpetual motion machine, it just cannot be done, perfection is impossible. So if the goal is to play perfectly “tight music,” then just stop now, because it will never be enough. But if Martin can just go into it for the journey of creation and engineering and brings the FUN back into it, we could really see something amazing. We just want to see something cool and hear Martin make music!

[@Eaode]

did you watch the video? Did you miss the part where he explains the "music video" he made of the machine was only 5% sounds made by the machine? He's obviously doing the technical work necessary to make a machine that can actually deliver what it appeared to deliver in the first place. It's weird to see the comment section saying this is somehow wrong-headed or cynical when I think it would be exactly that to just pretend that an asynchronous, barely-functioning percussion machine is actually "making the music" you heard in the video. Clearly this is a creator wanting to actualize what the internet thinks he already accomplished. The feeling of making a machine that makes that music, instead of a video where that appears to be the case, sounds like a much more satisfying goal as an artist.

[@BlackQuest575]

​@@Eaode ok so stay with me here go back and listen to the first original Marble machine video and tell me it sounds bad. the reason the MMX was only 5% was because he only enabled a small portion of instruments if the entire machine was on it would obviously be a higher %

[@d4slaimless]

@@BlackQuest575 Martin himself pointed out that first MM wasn't actually playing from start to finish. It didn't really work. The track was done in small parts and compiled with some things played by hand.

[@zefellowbud5970]

@@BlackQuest575 martin wants the marble machine to be able on its own without editing. i think its admirable

[@CristianKirk]

I really thought that we had already confirmed that marbles can play tight music. Oh, well... You know the deal.. Groundhog day is here.

[@logtrimmer]

He found the marbles themselves could, but not if his prototype could

[@trevorkeenan3512]

I would like to recommend a experimental statistical method called "Design of Experiments". It basically simplifies the experiment down to factors (variables like flywheel weight, counterweight weight, profile length, etc.) and the response (ms response deviation). You can then use these methodologies to find your optimal factor values to get the best desired response! It can seem a bit complex at first but is a genius way to run complex experiments like you are doing now. Just a suggestion!

[@_ratso]

The end goal was a world tour, utilizing a marble machine. Can you imagine being a member of Wintergatan, waiting to play this tour and then hearing that it may never happen because It can't be on time within x milliseconds?

[@br52685]

Do you *really* think the band is still together and just sitting around waiting?

[@UnknownUser-nz3io]

The end goal was to milk this channel for all its worth

[@wero5483]

I think this channel is the perfect example to what happens when you give unlimited time and room for improvement in a proyect to a person with obsesive traits. This will never be over, it will never be perfect enough for Martin.

[@dustybizzle1]

I think this comment section is the perfect example of youtube comments. Endless criticism, with no one jumping in to do it the way they want, but rather tearing down the one person actually trying.

[@justinnaramor6050]

@@dustybizzle1 Yes! Finally someone who actually understands what the real, true problem is with this fucking series. It's the winey commenters giving endless criticism. It's all bullshit.

[@bongi6811]

So this prototype plays more than 3x times tighter than you could by hand, even with a click pattern? That's awesome and surely good enough to play amazing music.

[@gutterg0d]

It needs to be as good as the expectation Martin has of an actual drummer though, not as good as Martin bangs the sticks.

[@gertjanvandamme2068]

exactly, at this point he only has to do finetuning like minimizing friction or flex in certain parts

[@mf_rat]

Then there'll be six months of collating data from various drummers. @@gutterg0d

[@gastonbell108]

Among drummers, to be called "tight" is an insult. It means robotic, overly precise, rigid when you should be flexible. Like a drum machine. Neil Peart, even after becoming famous, was dismayed by what a "tight" drummer he was, so he took lessons for several years and basically rebuilt his entire playing style to sound more organic and natural. Martin is chasing a monster that exists only in his own fears.

[@peronkop]

He needs to get it down to picoseconds.

[@80211Denver]

I love how he's trying to get to CRAZY levels of accuracy, has a breakdown, builds the most ghetto ratchet out of plywood chopped into shape with maybe a claw hammer as a tool and we're off.

[@br52685]

Agreed--absolutely clueless why he'd attempt to "measure tightness" on such a poorly-assembled contraption. But on the other hand, it's 100% expected for the trajectory of this project.

[@Steeger007]

I run sound for a few bands and played drums in the past. In my experience, drummers don't play "tight". It's more about the feeling and vibing with the band, music, and audience. They're not robots. Unless the song calls for "tight drums" they're going to play however they want. I've seen tons of circumstances of drummers accidentally speeding up, slowing down mid performance, and a lot of it has to do with how they're vibing. They might be able to keep the beat, but they also change it depending on circumstances, and are hardly perfect. If you asked them if they could drop a perfectly tight beat, I'm sure they could get close, but they're not concerned with how tight it is while playing. The only time I've personally tried to perform tight music is in a marching band, moreso in the military when it had to be perfect with 7 other drummers. I don't think that's what you're trying to accomplish here. For the sake of the marble machine, I know you want it to work properly and consistantly, but is "tight music" what you're actually trying to create during a live performance, or is it something else?

[@scaredyfish]

I think there are multiple levels of tightness. Consistency of tempo is one, and playing perfectly on beat at any particular moment is another. Both of these can be artistic choices, but the for the machine I'd say the latter is more important than the former. Tempo drift would have to be quite large to be noticeable, but too much variability within a short section would feel loose.

[@BlueHawkPictures17]

I think that if martin wants to "vibe" the beat in the way you describe, then we would first have to make it super tight, and then design functions that allow him to modify the beat in whatever way he'd like. In other words, I don't think the drummers in your story nor you nor Martin would want unintentional and uncontrollable beat changes.

[@gastonbell108]

Came here to say exactly the same thing. I know Martin is an accomplished musician who has toured with Wintergatan all over the world.... how does he not know that calling a drummer "tight" is an insult, not a compliment? How does he not understand the art of the percussionist? Does he think we are all merely music boxes of some kind, with little keys you can wind us up with? Neil Peart spent almost a decade re-training himself NOT to play "tight" AFTER he had already gotten famous and been declared a genius. Martin is going in the other direction, which is unfortunate because it's all in his head.

[@Steeger007]

​@@gastonbell108 Thats 100% what I'm saying. Perfection is fine if you're making electronic music as it's all computerized. However if you're playing live, you'll have your drums feel very mechanical and then you're trying to add instruments and human element on top of it and it's not going to mix correctly if its "tight" unless that's the type of music they're playing. I know he wants it to work correctly and be on time, but fractional amounts of deviation are hardly noticeable when playing live music. At some point, you're going to be too tight and it won't sound right. The machine can function as intended, but tightness doesn't account for the way something sounds unless that's the sound they're intending.

[@Eaode]

@@gastonbell108 this is the same argument artists use to say "I don't need to know 'perfect' anatomy, this is my art style!" perfection is not the goal of performance. But every drummer needs to know how to keep a steady and consistent beat before they can really "vibe with the band" at a high level. In the same way artists benefit from knowing fundamentals before they heavily stylize

[@314jake]

Jesus Christ Martin it's way beyond good enough at this point. You can be confident with (and proud of!) these results going forward

[@BrickBucketFilms]

I've noticed the comment sections since this third marble machine project has become akin to emotionally supporting a loved one who has become worryingly fixated on one aspect in their life and is letting it ruin their goals. Obviously this project is 100% up to Martin to do with as he pleases, but if the goal is to make music for others enjoyment I think he needs to consider he may be one of the only people that cares of a millisecond deviation at the end of the day. Even live performances by bands are far from that perfect benchmark. Should the goal in mind be a perfect marble based DAW, or should it be to create enjoyable music? Although not mutually exclusive, if the latter is more important, then I think it should be considered that the prior is not necessary. There should be a blind poll held to see if people really dislike or even notice between the tight and not tight marble music.

[@keiplaysgame]

the era of mmx was great. full of curiosity and wonder. now that mmx was gone, the only thing that i can see is a burnt out emptiness of passion. all of this will eventually come back to pointless restart all over again. Martin. i think if you just finished the mmx the way that the original did, it will be still be a great achievement. no matter how many flaws it has, it would eventually be an extraordinary piece of art. All those years just gone.

[@antiseth3964]

This feels like the strange birth of aperture science’s perpetual testing initiative lol.

[@russkuhn1354]

Such perfection is far beyond a human's capability to play . Milliseconds deviation? No one can even hear that difference

[@martindinner3621]

Part of the problem with small deviations is that if they are periodic and regular, they can cause harmonic effects that we CAN hear. (The "beats" that piano tuners use to get adjacent strings in lockstep.)

[@pomelo9518]

Beat frequencies are caused by out-of-tune instruments, not by instruments playing early or late.

[@NotLegato]

@@martindinner3621unless the marble machine is dropping along the lines of ten thousand marbles per second from two separate lanes, getting beating from it is impossible.

[@stonos282]

I think balance of the flywheel is one key factor to tight music. An easy way to test this would be to add weight at one side make it unbalanced. Keep up the good work!

[@John_Weiss]

The first thing Martin needs to do is _stop redefining "tightness"_ every time he comes up with a new test. The second thing is: decouple the Power Module from tempo control. If you have a plug-in metronome, you don't call up the power-plant and tell them to increase the frequency of the power lines from 50Hz to 75 Hz when you want to increase the tempo on your plug-in metronome. Yet that's _exactly_ what Martin keeps doing to the power-module.

[@Tsudico]

@@John_Weiss Yeah, if he wants to have "tight" music, he needs to stop considering the flywheel's tempo in the equation. Look at what mechanical solutions exist to variable speed governors so that he can choose the tempo via governor and see how tightly the marble drops are just from that.

[@jacobhollowell3087]

I hope he reads this, I was saying out loud that the fly wheel is bouncing slightly. And it is clear he needs to find a way to regulate the flywheel power. The flywheel should only be for power storage and not relied on for tight music.

[@hhurtez]

@@jacobhollowell3087 isnt his powerstoring the hanging weight, whereas the flyweel is inertia to regulate speed. But i think the Flywheel isnt the issue. Its the printed conections and his programming wheel. he should add sensor to his programming wheel.

[@sapiotone]

There isn't a wrong answer or bad idea so far in this thread!

[@juancarlosmarcos7401]

Listening that you still suffer from fails of the previous Marble Machines a quote came to my mind: Pain is temporary. Glory is Forever.

[@osculant]

Martin, I checked out the Speelklok museum the other day on your recommendation, made the trip from the Utrecht on my holiday just to check it out. Wish the MM1 was still there but it was still really inspiring!

[@Idiot354]

Is he dutch

[@doro5751]

Both the machines can now be found in Rüdesheim am Rhein, Germany. You want to check out @musikkabinett YT channel for some current footage.

[@IsaacDaBoatSloth]

the fact that the standard deviation of the marble system in its entirety is very close to just the marble gate proves its not worth dropping the project (yet)

[@joinslayer]

He doesn't know words such as precision or consistency. All he knows is tightness 😋

[@JensM-jc5dt]

There's been a lot of negativity surrounding all of this lately but nobody can deny that this project was amazing for all the time it ran and I'm very happy to have whitnessed it. Decisions are hard, and thus it's always more pleasant to hand them to somebody or something else, but in the end there is no right or wrong. I've had a really nice discussion with a friend lately about the book "The Midnight Library", and how we both felt like it wasn't as moving as it pretended to be. However I think it did a very good job in illustrating that no matter wich way you choose you'll always end up somewhere. There is no right or wrong. True, there is an exceptionally perfect version of the future in all those somewheres, but the chance that you'll end up in exactly that place is astronomically small. There are so many small decisions that influence your life all the time that it's almost pointless to seek perfection. So rather than chacing it and being scared of missing the turns, just going for it and seeing where you'll end up might be the more practical approach after all :)

[@UnManuco]

Have you tried slapping it and asking if its rushing or dragging?

[@MicahBBurke]

NOT MY TEMPO

[@monodiego86]

😂😂

[@killyGHILLIE]

😂 stiches

[@ulture]

I saw one of your videos that went viral years ago, had my mind blown, and subscribed. I've watched a few of your vids since then but haven't really been following the whole saga.... so I don't understand why you're so unhappy with it! What you've achieved is incredible, a thing of beauty. Technology and art, form and function, in perfect symmetry. I know that if you do choose to end this project whatever you come up with next will be just as impressive, but still I hope you don't end it. From where I'm sitting (admittedly I don't have a very good view), you're being way too hard on yourself.

[@nathnolt]

It's good that he's trying to optimize this little prototype, but using non final materials is definitely a factor, but I'm sure he's thought about that.

[@briancampbell179]

If anything, using metal parts to replace the 3D printed parts will be an improvement. He is basically testing a substandard version of the final product, so if the test stacks up, it is unlikely that the final one won't.

[@VacFink]

Perfection is the dominion of gods and fools. It is honorable to set it as a goal and folly to set it as a standard. If it were the standard it would unravel every creation by man or otherwise. I hope Martin can see the beauty and value in striving and failing to smaller and smaller degrees. The machine only wins if you stop.

[@Kheaz1]

Have you considered running 2-3 tests in a row with the same parameters to see if the results are constant?

[@Naugrimsilvertree]

That would be a great idea. Verify.

[@etcher6841]

Sound

[@Pistachio_mini_hotdog]

At this point I’m just convinced that Martin is trying to exhaust all options to convince us why he’s quitting the machine. It’s sad and I wish him luck, but I and many other people are gonna be upset when he finally shelves the project.

[@elnoggernog]

Always thought, building all that can never take so long. Since i started my diploma thesis i realized that it actually can. Huge respect for all your effort you put into this project

[@leerubin4303]

only takes this long if you allow project creep. This is actually bad engineering. The project will NEVER get done.

[@elnoggernog]

No Martin's gonna finish it someday. Even if it will take a 4th Version but he is definitely going to finish it

[@thomasbecker9676]

@@elnoggernog Historically, Martin has yet to finish a working, reliable marble machine.

[@noahstephens7622]

Do you actually play any instruments? It’s important.

[@giddycadet]

historically, martin has always let his perfectionism beat his ass. i don't think it's a problem with his engineering skills necessarily, unless you consider it an engineering skill - i feel like he was totally capable of the world tour with the mmx, but he simply refused to accept a single limitation.

[@badasson8825]

I think the loss in consistency from the higher BPM test may be due to the "prototype" itself. Although higher moment of inertia helps, the higher forces involved may have caused some unintended flexions in the parts and, with that, more backlash. Although, this is a good issue to have! It means the problem was not in the "design", but in the "medium". A sturdier prototype will probably increase consistency. As of clearance, I'm not criticizing the 3D printing setups. Much by the contrary, keep up with the rapid prototyping, there is more components to be validated before needing to make a more sturdy V2. Your work is one of the main reasons I went for a mechanical engineering graduation, thank you Martin for the great content as always!

[@Paul-D-Hoff]

Perfect timing makes music sound fake. It loses of human element.

[@tedferkin]

Failing early is the current strategy in a lot of software programming techniques. Although the full software often goes out, problems found, back round for another set of development. But hat is easier when you don't have to pull apart a mechanical beast to redevelop. But the "unit" test it a typical method of testing to get the right outputs from the right inputs. So if you stop the wrong inputs, to the next piece, it makes the next step easier to engineer

[@BrianHG.Ocean.Fitness]

You need to minimize friction to almost 0 on that programming wheel. Either that or make the friction a continuous level. (IE: make the programming pegs on the wheel all high and drip low to play the note, the immediately go back high for the rest of the time until the next note is played.) Also, if you wont place bearings on the lever which sits on the wheel, you may look into Teflon tape coating to make it as slippery as possible. Replacing the bike gear with an open wire an pulley system would be more fragile, but allow for much weaker spring load on the marble gate. Though, this means re-testing the marble gate timing.

[@RobertGarlinghouse]

You're building a clock. A lot of work reinventing something. Loved your first machine, hoped we would see many more marble machines over the years. I guess not.

[@theorganguy]

he never was on a path to create "many more" marble machines. really, he only wanted one. just that his previous attempts resulted in failures and thus needed a new version just to get to that "one" he can be satisfied with. what he is doing now can be more compared to the programming of computer games - he is developing a "new engine" - once that work has been completed, then anyone can go ahead and use that same system to build their own version and/or add more modules to his.

[@RobertGarlinghouse]

@@theorganguy he never will be satisfied. This project is moving further and further from all of the things that made the marble machine the joy that it was. He wants electronic precision, and by his own statements in this video, if he can't achieve that with marbles, he'll abandon the project. He doesn't admit it out loud, but he views the marbles and the mechanics to move them as the weak link.

[@jeffsucidlo9220]

I find it very interesting that in your quest to get well timed "tight" music, you are basically re-inventing a clock. You're even already using several clock elements like hanging weights and escapements. Don't give up. I think you're on to something.

[@Endless_Jaguar]

Perfect is the enemy of good enough, but good enough is a subjective metric. Rave on, rave on, until you can rave no more.

[@Sr7Sr7Sr7]

I wonder if the torsional rigidity of the bolts the gears are mounted to being so long is also affecting the timings significantly. Sounds incredible already though.

[@alltheeasynamesweregone]

Martin building the Marble Machine has been going for at least 4 years, when I first started viewing I was really excited to see Martin build a a marble machine and I planned to go see him live. It is apparent however, that it has become a channel for Martin to play around with different prototype contraptions and see if it can play “tight music” then make a small song and video..Stop, rinse, repeat. It’s been the same video for about a year. Sorry Martin, but it’s like constant foreplay with no pay off. Thanks for the content. Un-subbed.

[@euckb]

I've come back to this channel after a few years and honestly I'm not surprised it still isn't finished. It's almost like a mental illness. The perfection is annoying. It will never be finished. This is one of those mad artist dudes that dies with an unfinished creation because they went mad with perfection. Same videos, no finished product. Insanity.

[@Alexander.0671]

At this point the first song made on a new marble machine should be called "Tight Music"

[@ColtonRMagby]

The marble playing as tightly as they are this early in the prototyping phase is amazing. You've put a lot of thought into making this work, and it has paid off so far. Keep up the amazing work, the rewards will present themselves.

[@holographic_red]

Tightness in my opinion should be more about how tight all instruments in a track play together. So the grid can be a bit fluid, but the whole band plays by that grid.

[@noahjames]

Martin, I love your persuit for playing tight music. But you've already achieved i!. Those systems are working! The next thing you need to consider is controlling the tempo and making sure the machine doesn't slow down significantly when you read a lot of notes at once. Persuing music any tigher than that is just showing off at this point. There is a limit to human ability to percieve timing errors so small. I've really enjoyed watching your videos over the years and it's bonkers to see someone working on such a specific project for so long. Keep going and get it built! Don't wait for anything tighter!

[@nevernether3368]

As stressful as this project has been for you this is the highlight of my day getting to see someone pursue mechanical fun for a purpose. I absolutely love the marble machines and I hope you keep going with it. Amazing work as always man.

[@levilukeskytrekker]

+.

[@sjwimmel]

Two ideas: 1. A couple milliseconds is more than tight enough. 2. The real challenge will be that the drag will change depending on the song, because more instruments might be playing. For this reason I think a hand-cranked machine will be more feasible.

[@adamwarlock8263]

its so fascinating to see you applying the lessons learned! what a journey. keep it up, pal! you got this

[@levilukeskytrekker]

+.

[@thatguyStrike]

"Every week will be deeper testing to see how far I can optimize this prototype." Martin, I'm rooting for you to be successful and build the epic, perfectly working, super stable, transportable, simple and safe MM you are aspiring to create. But you "failed" (I still don't think you did tbh) because you rabbit holed, and frankly, you're doing it again. You're at what, one point something milliseconds (!!!) of tightness. That is incredible and will be way, way more than enough for live play or even recording (even 17ms would've been absolutely fine). Every video, the number of comments saying "please just move on and finish the thing, it's fine" is growing. I'm a perfectionist myself and I fully understand wanting to be better, better, and even better all the time. But for me personally, the time has come when it's just too much, because it feels like there's no self reflection here. I wish you luck and I'll tune in again in a while to see the progress, but I'm honestly too put off at this point to continue watching every week. You're amazing, Martin. An absolutely amazing content creator, engineer, musician (and hearing from someone who actually met you, a really cool and warm person, too). I wish you nothing, absolutely nothing but luck and success and I hope everyone who sticks around has a ton of fun watching. But this way isn't for me. Thanks for all the great times so far, looking forward to tuning in again at a later point!

[@CryptoTonight9393]

For the love of God and the accumulated centuries of time we have spent following these projects, do not abandon this as long as it can play as tight as a human. Dont rob the world of this masterpiece because of your perfectionism.

[@shaihulud4515]

I am not an engineer, but I've done a lot of similar projects, and I'd like to point out some issues that actually might help you: a) Do not make the same mistakes again! You have a much better approach in terms of planning and focussing on specific parts, like the marble divider, etc. But then there's "little" things you still leave up to chance, like arranging the distance between the two components with the axle. Your approach reminds me of "ah, it'll fit". Just don't! Measure those things carefully! b) The broken ratchet: this is a machine with a lot of interdependencies. So if you change one tiny component, you'll eventually get a great impact on another - always do backtracking, trying to identfy which parts are directly affected, and which might be indirectly affected. c) In your test setup you have a lot of movement. It's captured very well with the camera on close ups. There's a lot of rocking back and forth, apparently tiny movements, but believe me: they'll add up, and you cannot be sure as of now, in which ways they influence your testing results in terms of accuracy, reliability and validity. They might be miniscule right now, but on a bigger scale they might reveal themselves to be much bigger problems. d) This is more of a gut feeling, but...anyways: I'd test the machines' accuracy and tightness with a very low speed (like 60 bpm), a medium speed (at around 110 bpm) and at a high speed (at least 180 bpm). Just to be sure...

[@jaspermeuleman227]

Maybe you can add magnets to the flywheel and put copper coils on the side of it with a variable resistor, so you can tune the speed of the flywheel. Just like magnetic breaks on a rollercoaster. When the wheel goes too fast it automatically breaks. And when it doesn't it just leaves it.

[@nickgotvyak5890]

I think the mere mention of magnets will bring him Vietnam flashbacks...

[@shannonmurphy4914]

Windmill s use that same principle.

[@Herogaze]

good luck figuring it out. with all these parts together theres now a lot of variables. not that its impossible, but at least figuring out how to get predictable, consistent results seems like it would be important. Especially for improving the contraption.

[@iVehemence]

Never let perfection be the enemy of good.

[@thimomons4105]

It would be awesome to see how tight other mechanic instruments can play, because that is also something we think is tight enough

[@Runar_Tvalfager]

hey Martin, a thought experiment. have you tested to see if any LIVE stage has any deviation? cause i genuinely believe that any stage's sound system is not tight enough, so if the machines tight, but not the speakers, then the audience wont even hear the tightness to begin with :/

[@FitzChivalryFarseer2]

Holy shit I forgot this guy/channel/marble instrument existed. Nice to see you alive and well and still being brilliant at making your own instruments

[@kaasmeester5903]

Putting the marbles back in the last test looked a little like that I Love Lucy episode in the chocolate factory (except that you didn't have to eat excess marbles)

[@hashtagPoundsign]

Your work on the Marble Machine is absolutely mesmerizing, and I’ve been thinking about an idea inspired by an educational toy. What if you explored incorporating a mechanical ‘capacitor-like’ system using ratcheting springs? This could potentially help store and release energy in a controlled way, enhancing the precision of timing, especially under varying loads. Just a fun thought to consider in your amazing journey of crafting these intricate musical machines! Keep up the fantastic work!

[@Fsclh]

Since friction decrease precision, could we reduce it by adding a ball bearing at the end of the reader, on place of a resistive friction point ? Otherwise, I would expect an accumulation of imprecision proportionnal to the simultaneous notes on a beat..

[@stian5484]

Hi Martin, I am sort of embarrassed to say that I stopped watching you after you quit marble machine x. I thought you had lost the artistic touch with the project, but I was wrong. I started watching again, and I am proud to say that I'm really excited for this project. Go Martin!!! You definitely have not lost the artistic touch :)

[@beansnrice321]

Not gonna lie, I think a teeny tiny amount of deviation is good. It adds a little soul. =)

[@bitethebeat_music]

Thought the same few months ago. but someone told me: deviation must be less as possible. Then you can add this "soul" thing withing the trigger profiles.

[@Iangamebr]

Uncontrolled errors are not "soul" is just incompetence, the "soul" you are talking about is when a pro proactively choose to present something a certain way.

[@theorganguy]

@@Iangamebr I have to disagree; we are approaching this from the other end of the extreme: If everything plays exact, then it is entirely devoid of soul - it is just a very precise machine/robot - any and all deviations, irrelevant if wanted, controlled or uncontrolled, ADDs soul

[@Iangamebr]

@@theorganguy you have no scientific method to back up any of what you said. The greater the accuracy the greater the quality, that is quantifiably and objectively true, doesn't matter what your feelings on the matter are. The best product will, of course, be the more precise one.

[@theorganguy]

@@Iangamebr you are trying to quantify "soul" scientifically now? I'm not sure how to convince you. What I *do* have is experience - and I have experienced perfectly set music and it is: drab! boring! lifeless! ...perfect, yes, but not enjoyable. I do a lot of arrangement work and I'm fully on board with anybody else that has commented "here we are working on ADDING deviances to make it sound better and he is trying to get them to nil"

[@doggiegamer4464]

I started smiling the second he showed the first set of test results because that tells me that it's not going to go down without a fight. Never give up on this, you are doing something amazing here! :) Please never give up

[@BuddyCrotty]

This is already promising enough! It takes nearly one ms for sound to just get from one ear to the other, we really don't need sub 100ms accuracy for a live performance. I'd be interested to see what your normal live performance tightness is, or what you can even actually perceive. Just for some additional perspective, your prototype is already beating out the tightness of a Rolex (if not the accuracy outright)!

[@NickCombs]

Much of the best music ever recorded isn’t “tight” by Martin’s standard. It’s just not a good benchmark for one’s ability to make effective use of a musical instrument. A better question to ask would be, “can this instrument create emotional impact with the audience?” And I would venture to say making it more robotically precise takes you further from this goal.

[@oh_finks]

nice to see your collaboration with a metal band.

[@MicahBBurke]

What made the original MM so awesome was this rube goldberg contraption playing music. Not that it was tight, not that it was perfect but that it played an awesome tune by being wound up and held together with shoestrings and tape. The intent (I THOUGHT) of MMX was to make a similar machine that could go on tour, not to make a perfect computerized MIDI system that would never make a mistake. Awesome music was being played by a machine using marbles of inconsistent weight and inconsistent size through a system that was inherently inconsistent. The saddest part for me is all the amazing music we could have had. Dude has lost the plot?

[@roberthanulec5987]

I don’t care about tight music I just love the mechanics of it.

[@hunterdail]

its crazy how in the final test you managed to keep the tempo perfect while feeding the dropper handfuls of marbles, long live the marble machine

[@djtimo]

your tempo management is actually quite good. I think you could definitely play drums, even if it would just be simple beats

[@deathr1te]

Martin, Great work so far! I'm excited to see you continue exploring the possibilities for the Marble Machine projects, and the direction you've taken on the latest iteration of MM builds I fully believe has the greatest opportunity to achieve total success. Thank you for sharing your process of research, prototyping/testing, and development. As the individual pieces you've already designed begin to come together, I hope that the destiny of a future MM become clearer. If there is to be a future MM, I'm sure it will be the exact machine which calls to Martin, the Musician, to inspire him and enable him to create the music of your vision. Best of luck!

[@ScaldingCoffeeCup]

2.5 to 3ms of deviation sound like the sweet spot for me. Anything lower and we're going into the uncanny valley. In my songs i literally offset some drum midi notes so it sounds more realistic. I beg you, please don't abort that wonderful project because you wanted to sound too perfect. It's too precious

[@philipbechtle2828]

There is another important issue that probably has not been mentioned yet: the effective music tempo is a complex balance of resistance from the machine, governor settings and weight in the Huygens drive. You might be able to tune the weight against the governor for fixed machine resistance to achieve a certain desired tempo. However, already here, your test indicates that this is a fragile balance. Even more critical, the machine resistance will change in different phases of your songs, when parts of the machine are switched on or off. This will change tempo. I would guess that on-the-fly adjustments of the governor are going to be difficult. This could ask for a much more complex governor design, which is an entirely new rabbit hole.

[@EricTrimbur]

Tell us Martin, is this "Tight Music" in the room with us right now?

[@EricTrimbur]

Seriously though, if you wake up tomorrow and realize this whole tight music thing is silly we will all be overjoyed. I understand having tunnel vision like that. The best time to wake up is now. And the second best time is tomorrow, just don't let it go too long. We're rooting for you buddy!

[@AntimaterieGame]

Don't be too hard on yourself Martin. I remember you did some prototypes on the MMX too. If nothing else the machine looked awesome and so many people enjoyed watching you build it!