Rebuild? Why? - Marble Machine X #142


All wintergatan Music is now available for background music use in videos and livestreams!
Check out the free License Here:
A lot of REAL progress this week! I thought it was high time that i tried to give you all some context on why i am doing this major redesign that i am working on right now. Im presenting one theory that might explain the motivation behind the decision to do this redesign. Perhaps the Marble Machine X itself is the Local Maximum, in that case, the current process will lead me to that discovery and i will take another u-turn to find the path to the world tour. Probably i will take the advice that has been given to me so many times during the past years: "Why dont you just play guitar?"
:) Hope you all are are having a great day and enjoy everyone! ;) / Martin

Rebuild? Why? - Marble Machine X #142
Video edited by Martin & Hannes Knutsson
Support the Marble Machine X Project:
► nolocal.infojoin
#marble #machine #4K
INSTAGRAM ► wintergatan2000
Thanks to the Marble Machine X Team 2020
►Alex Füßl | Alex CNC | Munich, Germany
►Carlos Montoro | Anjuda Guitars | Madrid, Spain
►Chris Nadon | Aix-en-Provence France
►Dwayne Myers | Safety | Phoenixville, PA, USA
►Ecco Pierce | All Things Ecco | Oakland, California, USA
►Florian Hu | fhuable | Wellington, NZ
►Jacob LaRocca | Rocket Props | Somerville, Massachusetts
►James Passmore | CAD9 Design LLC | Worcester MA
►Marius Hirn | Munich, Germany
►Nick Householder | Florida, USA
►Philip Brown (solo banjo) Winston Salem, NC USA
►Rainfall projects
►Richard Southall | String HeArt | Devon, UK
►Sebastiaan Jansen | Leuven, Belgium (BE)
►Tim Keller | Inventopia | Davis California USA
►Tim Mitchell | CNC Machinist | Nashville USA
►Tobias Gutmann | Tobias Gutmann Prototyping | Munich, Germany
►Tommaso D'Amico | Vancouver, Canada
►Will Francis | Machine Thinking | San Francisco Bay Area, USA
►Blue Sparkle and her great team for this video's subtitles (and many others).
►Grant Lansdell: Great Idea Provider ;)
►Viktor Stenberg, Calle Guldstrand, Gustaf Törner | ProtoCut
►Liam Ward
►David Lewis
►Tobias Smidebrant
►Erik Holke
►CNC Routerparts
►Nathan Skalsky
►Karin & Olof Eneroth
►Matterhackers -
►This Old Tony -
►Thomas Pilot -
►Roys Sheds Perth Australia
► -
Help us caption & translate this video!


  • DoodleChaos
    DoodleChaos12 dager siden

    Ooof that was a brutal angle grinder. Hang in there Martin, regardless of what happens we support you.

  • SID


    2 dager siden

    You’re just dragging it out for the $27,000 patreon bucks every month.

  • TheChipmunk2008


    6 dager siden

    @Wintergatan YASSSSSSSS

  • JoRasEyez


    7 dager siden

    Redesign?????? am gonna stop watching i sad a million times Martin has a problem ... i wasted like 3 years following this project ........we all know martin he is perfectionist ,but every thing has its limit.. and i think he is doing this because the channel was not for building the marble machine , so he is scraping the barrel , soo it was fun while it lasted .... bye bye

  • numfortably_combed


    8 dager siden

    @Wintergatan Hi Martin, You're now reaching epic levels of failure rates with the grinder! ;-) People in the comments below are talking about 6 Sigma standards. I'd also mention that you've reached the stratosphere-- you're now talking about aeronautical design levels of failure. See also ideas about "redundancy" in design and "fault tolerance." Airplane safety statistics might also be helpful to get a grasp of how to pinpoint design durability and weakness. Keep striving for excellence!

  • Atomic Shrimp

    Atomic Shrimp

    10 dager siden

    If you had to approach a completely different, but equally complex machine build project in the future, knowing what you have learned along the way, what would you do differently? Seems to me that you might invest more time testing single prototypes of a sub-mechanism before iterating them across the machine itself - that is, for example, build one marble dropper, and test it with a million marble drops. Then you're only fixing one thing, rather than fixing 20 of the same thing, only to find that the fix isn't sufficient. I'm not saying I would do better myself (in fact I know I wouldn't - I simply lack the patience for this scale and complexity of engineering), and it's also less exciting and less fulfilling to be just testing components, instead of seeing the machine emerge of course. Hope this doesn't come across as critique - I'm actually just interested to know how your methodologies have changed across the course of the project.

  • Arthur Fleck
    Arthur Fleck2 timer siden

    YOUR 🧠 YOUR INGENUITY.. 💡 YOUR AMAZING MAN! AN INSPIRATION!!! 🤙🏽🤙🏽 MAD RESPECT! Thank you for sharing with us!

  • Wolf Hound bossy
    Wolf Hound bossy3 timer siden

    Rrrrrrrreee make

  • -Arkaiid-
    -Arkaiid-9 timer siden

    do a giveaway on the marble arms! your fans would love this, and it would e a way to get rid of them if you aren't using them- some kind of givaway/competition. i got this idea when you said "statue"

  • Linus Eklund Adolphson
    Linus Eklund Adolphson9 timer siden

    It’s NOT Sunken Cost Fallacy. Setbacks and detours are a natural part of the path and process towards any goal. This is a turnaround, not a failure. It’s a necessity for further progression and improvement. The MMX project can’t fail, as it has no definitive parameters for success - it can only be accomplished to a higher or lesser degree in so much more or less time. Of course you should continue. You, me and all of us look forward to the next step in the story as much as the finalization of the project with a world concert tour. We love to share your enthusiasm and despair, your aspiration and frustration, your elated victories and confounded disappointments. Look at it this way: it has been, and still is, a most glorious journey so far - why stop now? It’s all a labour of love and as long as you’re happy, we’re happy.

  • Homsan Lolisan
    Homsan Lolisan11 timer siden

    10:36 I'm going to try to appl.. oy, wow (LOL)

  • Jia Ern Chong
    Jia Ern Chong22 timer siden

    I'm very look forward to the success of the marble machine x. It's more hardwork to rebuild it but I will support you anyway. The melody of the old marble machine still is the favourite of mine. MMX undoubtedly will be much more better.

  • gnarly cool dudes
    gnarly cool dudesDag siden

    why did you do this

  • Odyseusz Koskiniotis
    Odyseusz KoskiniotisDag siden

    Holy shit, please keep going

  • Random Voice ie. Ian Tan
    Random Voice ie. Ian TanDag siden

    Although the chronological progress is missing, I would love to see the excess plywood boards side by side with their intended design in an exhibition when MMX is finally done. It'd be like another self-playing music machine museum, this time, featuring the creator in the flesh.

  • BrianRDub
    BrianRDubDag siden

    The year 2023. "We are on rebuild #37 of the MMX!"...... I think this is the strategy...It never really gets done. Only improved perpetually. In the real world this is called scope creep and has brought many a good project to its knees and onward into failure. Here in youtube land there is no such thing as scope creep or failure. Only excited supporters watching the evolution of something fascinating.

  • momentinpassing
    momentinpassing2 dager siden

    This is why the human mind has so many stabilizing features. Falling off .01% of cliffs, or eating .01% of the broken glass we find, is rubbish.

  • Kalle Jillheden
    Kalle Jillheden2 dager siden

    Hahaha that sick illuminati-esque graphs comparison, love it

  • Coxy's Railroading
    Coxy's Railroading2 dager siden

    Paste wax is also good for reducing friction on wood surfaces.

  • Coxy's Railroading
    Coxy's Railroading2 dager siden

    Discovery & Analysis => Improvement & learning => Progress ^____________________________________________________I It's a complex project without a robust precedent. Iteration is the way forward.

  • KingPatrick 12
    KingPatrick 122 dager siden

    Hey man i remember when you came out with that video in 2016 and I loved that video then. I rediscovered it and I realized that one day this machine will be in a museum. If you need to take a break at anytime we'll understand.

  • Dan
    Dan2 dager siden

    idk man i think youve entered a black hole

  • hypothetical their clavicle
    hypothetical their clavicle2 dager siden

    If you're worried about design flaw you can't have metal on wood friction. Might as well change now.

  • DarkhFyre
    DarkhFyre3 dager siden

    this is honestly such an incredible project. making a machine that plays music isn't an easy feat and it is amazing to see you work on this project and refine it to be a music and engineering masterpiece.

  • samili
    samili3 dager siden

    Why don't you add another marblegate per channel and reduce the gear ratio, that will give each marble double the time to get settled in the marblegate. You'll have to slim down the design tho to keep the width of the original design

  • Marco Roose
    Marco Roose3 dager siden

    Maybe you could try „Silbergleit“ instead of Graphite Powder. Its more like a paste. I‘m really amezed every time when I‘m using it how much it reduces friction.

  • Daskut
    Daskut3 dager siden

    "I'm rebuilding the Marble Machine X!" Germans: 24:49

  • Pixel The Incrediboxer
    Pixel The Incrediboxer3 dager siden

    This is *marbelous*

  • Joe Balint
    Joe Balint3 dager siden

    Martin- I've missed quite a few episodes. But, I had a thought after your Captain Context vid: Are you treating the wooden parts to reduce the impact of humidity? A couple of decades ago, we use to make our robot chassis from plywood. We did our work in Ohio but the competition was at EPCOT in Florida. To avoid the plywood swelling from absorbing water due to the high humidity, we painted the chassis with several coats. Is there a coating on your plywood parts? Maybe it won't be an issue with your smaller parts.

  • lexdimond1
    lexdimond13 dager siden

    You need to link up with Adam Savage

  • Benjamin Stadler
    Benjamin Stadler3 dager siden

    This calls for an intervention. Somebody get him out of the shed? Feels like current circumstances take a heavy toll on artists. Nothing wrong with focusing on details but this turns more and more into an unhealthy obsession - called perfecting yourself to death. It's not worth going insane over this.

  • KWA
    KWA3 dager siden

    If it were not for this kind of ambition and determination the world would never have the light bulb, the automobile, or the airplane. Keep up the good work.

  • Hadley Siqueira
    Hadley Siqueira3 dager siden

    I feel sorry for those ones who are making the 3D model of MMX. ^^'

  • PurpleGuy2779
    PurpleGuy27794 dager siden

    when i hear brutal, i think of sledgehammers.

  • Evan Stunt
    Evan Stunt4 dager siden

    Someday you'll need some magnetic polarized balls repelling to each other to avoid some noise .. or mechanism .. try adding alternated magnet binaries .. if they repels each other there is no reasonable motives to the marble to get stocked .. cause it will flow by magnet repulsion almoust for about forever..

  • ronrooz
    ronrooz4 dager siden

    Martin, why is it that in many of your videos the sound is sometimes out of phase and a moment later (next edit step) back in phase again?

  • Uraneum
    Uraneum4 dager siden


  • Baron Münchhausen
    Baron Münchhausen4 dager siden

    Why? YT-Money for this "never ending story"!

  • Anson Olds
    Anson Olds4 dager siden

    Piano Technician here, I would advise against using Graphite Powder. In pianos it is useful for the short term, but eventually it gums up. Some "technicians" use graphite so much that it ends up everywhere, and those parts need to be cleaned or replaced because the friction becomes unmanageable. Instead I would recommend using Teflon powder, very sparingly.

  • Abe Dillon
    Abe Dillon4 dager siden

    An alternative to graphite is High-Density Polyethylene (HDPE) . It's like a low-friction plastic material that should be pretty easy to machine.

  • Abe Dillon
    Abe Dillon4 dager siden

    At some point, you gotta lock in the design (flaws and all) and push future improvements to MMX 2.0 After the world tour, you'll have money for all sorts of fun tools.

  • Willaby Uberton
    Willaby Uberton4 dager siden

    I think your math is off. If you do something 100 times with a 99% success rate, than you have 0.99^100 = 0.366 chance of it working. Obviously not good enough, but that's still not 100% chance of one failure. A 99.99% success rate 3600 times is 69.76%. Increasing success rate by a single nine gives you 96.46%! If you want to play the machine ten times a day with a >95% success rate (less than one failure a month), you need each individual operation to have a 99.9999% success rate. Which is probably beyond standard engineering. NASA quotes figures like 99.74% success rates for engines. Have you considered finding ways to recover gracefully from a failure?

  • Ethan
    Ethan4 dager siden

    me: doesn't watch wintergatan for a few weeks, come back. martin: I've replaced half of the machine. Wow what an improvement. The lack of noise on that runthrough :o it's beautiful

  • MantraHerbInchSin
    MantraHerbInchSin4 dager siden

    He is not crazy, just devoted. I support you Martin!

  • Martin A.
    Martin A.4 dager siden

    yeah... lets start all over again.... at this point this is so stupid

  • Jörg Johow
    Jörg Johow4 dager siden

    I changed my mind:"never" - period.

  • Carson Kendall
    Carson Kendall4 dager siden

    you have the coolest job ever.. if you can call it that. incredible.

  • Somebody_Else
    Somebody_Else4 dager siden

    15:38 Good spot, Martin!

  • teamidris
    teamidris4 dager siden

    I hear your ‘feels’ 100% :o) I ran the same race car 4x4 from 1993 to 2020 and it got changed many times. Around 2011 I took an angle grinder to the back and was cutting out redundant brackets from 1993. It was bad/good feeling. It was waaay better for it. *At the moment I am making a Mamod steam car into a Gscale rail car. I will have a small box of prototype parts that each took hours to make. :O

  • LangyX
    LangyX4 dager siden

    @wintergatan, Auction off your spare parts as souvenirs.

  • YourResidentRedleg
    YourResidentRedleg4 dager siden

    Not gonna lie. I don’t care about any sort of reason. I’m not gonna stick around for someone to rebuild the same thing over and over and over and over again. Been here for years now. Watching you do this so many times is just monumentally disappointing. Keep screwing yourself, the people who help you, and your fans. Just finish the machine dude. Probability and what actually will happen can be two entirely different things. But oh well. Just milk a Patreon for years with no finished product.

  • Frank Wells
    Frank Wells4 dager siden

    9146 X 5 = 45,730

  • Jonas Pihl
    Jonas Pihl4 dager siden

    I Can hear the cad team screaming

  • Jamie Kawabata
    Jamie Kawabata4 dager siden

    "All my friends have at least one friend like that." Hmmm... checks out.

  • wilson langeani
    wilson langeani4 dager siden

    You build it, man!

  • Ro Paul
    Ro Paul4 dager siden

    This is a prime example of why I unsubscribed from this channel. What will you redesign next? You can spend the rest of your life on continuous improvements. I can see it now Episode #507 - Marble Divider Version 8

  • Chuck D
    Chuck D5 dager siden

    I wish I was this brilliant.

  • Rafael Pernil
    Rafael Pernil5 dager siden

    I think it would be better to design the MMX to be serviceable, to have easy to replace parts and ways to recover from failure quickly

  • Niko Shropshire
    Niko Shropshire5 dager siden

    I am not sure if you want to get your machine complete. Wth?

  • Skillfully Executed
    Skillfully Executed5 dager siden

    Well your probability math is a bit poor there but okay, u can't really say u have a 99% stability rate based on 100 marbles, that is not a dynamic probability. What you are looking for is stability, meaning a 99.9% probability for each marble to go through the system successfully, if after this you can start calculating using the probability of instance 1, and put the number of attempts as a power of. Probability of a marble stopping within 3584 time(s) at 99.9% probability = 0.9993584 = 0.027714675751556 = 2.77% So at 99.9% you'd actually have a 2.77% failure rate for each time you played the song, that is more than acceptable.

  • MC Falkenstein
    MC Falkenstein5 dager siden

    25:43 "I know my friends have at least one friend like that" That one cracked me up :D

  • Diário de um casal viking
    Diário de um casal viking5 dager siden

    4:35 Hey guys, ChrisFix here

  • Charlie
    Charlie5 dager siden

    Lolz that graph about the local max is hilarious

  • Charlie
    Charlie5 dager siden

    Your transparency is so refreshing Martin. Been a fan for quite a while and love how much detail you give all us fans. Cheers and Stay Safe.

  • nursefuzzywuzzy
    nursefuzzywuzzy5 dager siden

    Martin Don't you dare quit on this! You can get it to work I know you can do it!!!!

  • MeMuX
    MeMuX5 dager siden

    You should make a "Loctite usage" counter next video of assembly lol xD

  • AlexTheFluffyBoi
    AlexTheFluffyBoi5 dager siden

    I wonder if we'll get a remake of the marble machine song on the new marble machine i really hope so :3

  • Pete Hyde
    Pete Hyde5 dager siden

    Service with a smile

  • Michael Nio
    Michael Nio5 dager siden

    What program do you use for your cnc router?

  • 風見隼人
    風見隼人5 dager siden

    I hope I can see my name next time. QAQ

  • Candra Shalahuddin
    Candra Shalahuddin5 dager siden

    I have followed your progress in making this masterpiece. I'm very inspired! But at this point it looks like you are too emotional to do this brutal angle grinder. If you discuss this first on NOlocal before act, I will advise you to make an MMX version, finish your work in stages, don't destroy what you have previously made, make a new one that is better without destroying the old one. nothing is perfect, perfect is respecting imperfection But I respect your decision. Just don't get discouraged Martin. I hope the best for you!

  • Simorenarium
    Simorenarium5 dager siden

    As a Software Developer I can really relate to the complexity and sometimes to the necessety to redisgn things...

  • HackdIluminattihater
    HackdIluminattihater5 dager siden

    24:49 Nein! Nein! Nein! Nein! Nein!

  • Th3f0xman Mc
    Th3f0xman Mc5 dager siden

    martin the 371 thousand people that watched this video are the real supporters

  • G erman C arrasco
    G erman C arrasco5 dager siden

    In other words, songs for the tour are not ready, building life in France is cozy and COVID allows to postpone completition undefinetey

  • NaiOni
    NaiOni5 dager siden

    So this is what I'm wondering... Isn't the likelihood of a master musician missing a note greater than 0.00001%? And isn't the likelihood of a string failing on a guitar greater as well? It's starting to feel like you're chasing precision that might actually be unachievable. The Lego company has some of the most accurate and precise manufacturing of any factory in the world, and 18 out of 1 000 000 pieces are faulty. That's 0.0018%, that's a lower precision than you are aiming for with the MMX. You are building a machine at least partially by hand with the goal of it having higher precision than some of the most sophisticated machines in the world. I'm worried that your unwillingness to compromise could make the completion of the MMX impossible. I admire the dedication, but this is the first time watching these videos that I've truly felt that you might be aiming towards the unachievable.

  • Tedd
    Tedd5 dager siden

    As a comparison, how did the original marble machine fare stability-wise? I know a lot of marbles dropped to the floor, as can be seen in the video, but how much work was it to get one song out of the machine?

  • Tausif ali
    Tausif ali5 dager siden

    No one..... German Engineering: Lets make a band using marbles

  • Koawei
    Koawei5 dager siden

    classic to steampunk

  • Daniel Brendemuehl
    Daniel Brendemuehl5 dager siden

    With the graph Martin I just got to say I completely support you, but as a machinist I know that if you are dealing with anything mechanical you are going to have failures. Nothing will be perfect you can get close but nothing in this world is perfect.

  • chatparesseux
    chatparesseux5 dager siden

    What you described is 6 sigma standard there are training books to reach that standard in the engineering process. Most automotive industries follow it

  • Ev3rM0r3
    Ev3rM0r35 dager siden

    One of the most dedicated individuals I have never met but the most intriguing single person I've watched on youtube or anywhere. I've watched your progress from day one on the last build and now this new one. What a journey. I could never follow in your ambition footsteps. I envy you sir. And GL on completion. And god damnit you better complete this!

  • Forge
    Forge5 dager siden

    Welcome to the 5 9's club!

  • Rai Kanegasi
    Rai Kanegasi5 dager siden

    7:58 If you were going to bore me, I wouldn't be watching you. Keep going, you're doing great.

  • zerohm m
    zerohm m5 dager siden

    been following this for what feels like years now. really cool to watch the different iterations and major breakthroughs you have periodically.

  • ky
    ky5 dager siden

    At this rate I don’t know if it will ever get finished.

  • vinasu maaj
    vinasu maaj5 dager siden

    "In mathematics you don't understand things. You just get used to them." - John von Neumann

  • Tony C
    Tony C6 dager siden

    You're kidding me.... just finish the machine already!

  • Damalycus
    Damalycus6 dager siden


  • vinasu maaj

    vinasu maaj

    5 dager siden

    the last part about stability hits home as a software engineer

  • MrCyssiek
    MrCyssiek6 dager siden

    Martin... I stopped watching this amazing project because... you tripping dude. You should know where is healthy line because MMX is not your project anymore, its obsession.

  • Drew
    Drew6 dager siden

    Martin: takes out angle grinder Cad team: NONONONONONONONONONONONONO

  • Jakubiak Kos
    Jakubiak Kos6 dager siden

    20:00 Explanation

  • Steven Kozlosky
    Steven Kozlosky6 dager siden

    I love what you've done and what you're doing!! It's so much more than a build - it's instructive. My son is currently embroiled in a senior project at university. It's a device that four students conjured up in a spitball session and now they're tasked with building it. It has electronic and mechanical components but I'm convinced that the lesson is elsewhere - the lesson is the human components and learning to usher resources. The likelihood of a device that four kids agreed to build based on an hour spitballing ideas will work is slim to none and not really important. I concluded this having watched our eldest go through a similar "senior project" at another university. It was, by my estimate, not successful as a product, but what his team learned about engineering and management was priceless. Thanks again for all you do!!!

  • Carter Laubach
    Carter Laubach6 dager siden

    Don’t give up! You have made too much progress to turn back!

  • Omnihil777
    Omnihil7776 dager siden

    November 2036 : 6th rebuild of the MMX on the way, we think it will be ready in 2 months...

  • Exploding Crow
    Exploding Crow6 dager siden

    this whole project is proof that engineering and art can coexist also i'm from chicago, i have no idea if it's actually a 'chicago' or 'boston' screw, but apology accepted anyway also also, if you look briefly at the history of machines moving really fast, before the industrial revolution there was no real concept of machinery failing through normal use. things didn't move fast, you didn't have a part that was being used hundreds if not thousands of times a second. so when things started breaking on the big expensive new machines it took a collective brain blast from engineers to figure out what was happening. you basically went through all of that independently. i went to engineering school, i'm not even sure if that's where i learned about the problem. if it was they mentioned it like once and never again. the biggest breakthroughs of the past have become so common that they aren't even considered important enough to be common knowledge

  • Jasper Cattoor
    Jasper Cattoor6 dager siden

    Noooo don't give up :o

  • Jasper Cattoor
    Jasper Cattoor6 dager siden

    Yes, exactly, I mean WTH? WHYYYYY

  • nur farrahana
    nur farrahana6 dager siden

    You are so geneous

  • Shankara Rhuddlan
    Shankara Rhuddlan6 dager siden

    Hi Martin! Before I say anything else please know how much we appreciate you and how much you have inspired so many of us!!! I had this idea... what if you take any of the parts you removed and won't use again and auction off some of these parts to help raise money for upgrades, stir up some hype and excitement and allow some of us to donate and possible win a piece of MMX history and display it in our homes while we continue with you on this journey? Thanks for reading my comment and thank you for being so awesome!!!!

  • CarColleRote
    CarColleRote6 dager siden


  • soniyu ziuy
    soniyu ziuy6 dager siden


  • soniyu ziuy

    soniyu ziuy

    6 dager siden

    ... will last months. The difference is remarkable :)

  • Sean Lyttle
    Sean Lyttle6 dager siden

    6:12 - Thumbs up for a perfect transition and a perfect illustration of your point!!! Nicely done.

  • Jacob Johansen
    Jacob Johansen6 dager siden

    the last part about stability hits home as a software engineer

  • Jacob Johansen
    Jacob Johansen6 dager siden

    Teflon shims could also work as a friction reducer without the need to reapply the graphite lubricant over time... its what's on the bottom of mice