Same pattern in software. The best tools for specific professions usually come from someone who worked in that profession first, got frustrated with generic solutions, then built exactly what they needed.
General-purpose tools try to serve everyone and end up with features that kinda work for most cases. Someone deep in a niche builds for the edge cases they actually encounter, and it turns out those edge cases are universal within that niche - they just weren't visible to outsiders.
I attempted to productize a super niche idea I had for myself that I have ended up using multiple times. It's a simpler method of running wires without having to cut/patch drywall and drill through studs.
It started with trying to create a router-like device equipped with a motor (which was version 1 [1]).
But then iterated it to be a plate which can be attached to existing routers. That resulted in a drastically improved user experience which went from doing it all inside and making a mess to doing it in your garage with more precision. I've used the final design on 4+ projects at home and it has saved me so much time [2].
CAD is complicated, yes.
But the biggest pain point is that engineering requires a lot of adjacent knowledge about material, tolerance, mechanical design, tooling, and so on.
Making things requires patience.
That really depends on what you're making. I've made a lot of things that are basically, I want an object with this shape, model that shape, print that shape, success. For things where tolerances and material properties matter, a little trial and error takes care of a lot of it.
There's an engineering saying that anybody can design a bridge that won't fall down, but it takes an engineer to design a bridge that just barely won't fall down. Why do you want a bridge that just barely won't fall down? Because it's a lot cheaper to build. That's not much of a concern when you're printing little doodads at home. I waste some material by designing overly-strong structures, or getting it wrong and iterating. That's fine, the stuff is cheap.
My wife and I bought a Bambu P1S and I think the machine has had a duty cycle of 30% (excluding when we weren’t in town). It’s great fun. The majority of the models are ones we get from the Internet, it’s true.
But my wife used some base open source components to design a block that goes into the bits of a playpen that we used to have and transform it into something that docks with the wall instead of only with itself[0].
And I have designed with Claude a few small things like card holders for the board game power grid[1].
I wish there were better AI tools for interacting with modeling software. As it stands I use OpenSCAD with Claude and that seems as good as it can be. There are Solidworks AI startups but they’re like for professionals.
The Bambu P1S I have is quite low friction to set up. And I have an AMS2 Pro on top of it that feeds different kinds of filament (material and color) into the printer. I have just the one but now I wish I had more AMS hooked up.
AI coding tools are providing non-programmers giving similar ability. Not production quality, but still useful to them everyday and they can tweak as they go.
Making a good mechanical design is difficult, it usually involves making several iterations which in the physical world takes a lot more effort than iterating on software.
One thing that always bothered me is that most people with 3d printers seem to design things on their own from scratch and rarely take on others designs, improve them and share them. There is little collaboration going on for 3d prints in comparison to software. Except from maybe ~10 widely successful projects that now have healthy communities improving them.
Why is GitHub and similar sites not used more among makers?
You're assuming that CAD tooling is mature enough to enable that.
There is no standard format for CAD projects/design files. STEP is a standard format for exporting finished designs.
There are many CAD-with-code platforms, but none of them converged on a shared language the same way there are multiple competing C compilers.
I might be beating a dead horse with this one, but standardizing around FreeCAD isn't possible either, because it isn't good enough to compete with commercial CAD software like Solidworks, OnShape or Autodesk Fusion. Blender is almost there though, but for mesh-based free form 3d modeling.
Then on top of that, a 3d model on its own is useless. You need to manufacture the part. Printers have varying capabilities and something that has turned out to be particularly essential is multi color printing, since it allows embedding text and markings onto a print but not every printer is capable of doing it or doing it economically.
Ordering individual parts is expensive, which means you'd rather buy a full kit from someone who is getting volume discounts.
Wat? Android, Linux, BSD (incl iOS and Mac) absolutely dominate the market in terms of mind share and deployment. I can't think of another philosophy that is so directly, causally responsible for billions of dollars of realized value.
It sounds like you're parroting the corporate line of the early 80s. "Making money directly selling software artifacts is the only way to win." Which, as we know in retrospect, was a completely failed strategy, steamrolled by companies which... wait for it... adopted more flexible technology based on the Unix philosophy.
I'm going to lump together proprietary and for-profit software a little because I rarely see exceptions:
Proprietary / for-profit software is made that way, yes. There's no other way you can do it, given the incentive landscape. You take the huge base of FOSS code and spend as little effort as possible to build a thin little layer of moss on top. You respect the customers as little as possible to keep them around. You fix bugs as little as possible. You use other proprietary services that spy on your customers, like Sentry and Firebase, because privacy costs money.
Free / libre software has to compete on its merits. If a project isn't useful it doesn't grow. But not growing is okay. Some projects accrete over years, like a pile of stones forming a cairn. They don't need to squeeze money out of people to live because they aren't alive. They don't have to "eat".
I'm mixed on the Unix philosophy. It makes a lot of sense when you're building CLI tools that a hacker is going to plug together, because then your program is really a function in a programming environment that spans one or more entire computers. Tools like ripgrep, jq, curl, they're all great, I love them. A good function does one thing and does it well.
But just as often, I'm okay with huge software that does a lot. Web browser engines are evolving into universal GUI / IO frameworks and I'm trying to make peace with that. Systemd does a ton of stuff, but hell, so does the Linux kernel. I don't see an inherent problem with having an init system that acts like a monolith kernel for userspace. Microkernels are nifty but in the end we all ship our org charts. Maybe there's no need for microkernels if the real divide is between kernel programming and userspace programming. For the same reason, I haven't found any personal use for wasm, because wasm makes the most sense when you're connecting two pieces of code written by different teams at different times, like a GIMP plugin. I don't need wasm to plug my own code into my own code.
And for GUIs it's just been a fucking nightmare. 57 years since the Mother Of All Demos and it's still 10x easier to write `fn main()` and build a CLI program that runs on _every_ OS, rather than a GUI program that maybe runs on one OS, like it runs on Ubuntu 24.04 but not Ubuntu 22.04. What a fucking mess. GUIs don't compose, so every GUI project I've tried feels like I'm inventing the universe from scratch. It's fun but it's a stupid fucking waste of time.
Honestly browser engine frameworks like Electron and Tauri _are_ the Unix philosophy compromise. Making a GUI framework requires tens of thousands of lines of high-effort code made by experts. If a browser's one thing is "Be a GUI framework" then it allows your GUI app to just serve HTML and now it's a CLI app that runs anywhere without fucking with GTK 3 vs GTK 4 bullshit.
Sorry for the long rant. This stuff is all percolating in my head. I started with Visual Basic 6 and I still haven't seen GUIs improve since then. Phones have been a fucking step backwards, too. Everyone uses a phone but just like a mouse-and-keyboard player dominating gamepad players in a shooter game, I get a lot more done at a real desktop with a real pointing device.
I've been telling myself that for as long as 3D-printing has been consumer tech (about 20 years ?) and now it's shifted to "I'll borrow one my friends' printers if needs be".
In truth every time an issue fit for 3D printing has come up in my life, I solved it easily with wood and cardboard. I'm starting to recognize I might be a craftsman at heart.
I have had a similar experience; my preferred material to work with is wood. However, as I got more into tinkering with electronics and vintage computing, I'm finding more instances where wood does not achieve sufficient strength-to-weight ratio, especially for small parts where wood grain and anisotropy becomes a significant factor to consider.
Borrowing from a friend/library/work/low-cost maker space is the way to go unless you plan on printing with the thing for numerous hours per day on average. Having said that, once you start 3d printing, it becomes a tool you reach to more and more
Check your local library, a lot of them have one and can help you get started. It's usually pennies to print too. I printed an adapter for my coffee grinder at my local library a few weeks ago and it took 2 days and cost me $4. Fantastic stuff.
A side/tangent note, I wonder if it says something about myself/character, I choose the faster print than the surface finish, so I'll have this rough surface finish due to the supports. But it goes from 20hrs+ to 13.5hrs so it's like I'll take the faster option. What I printed was a shell with internal screw mounts, it was a big piece about 5.5x4x2.5" with 20% infill, regular temp/speed and the structure support for overhang.
It's actually a 4D printer unless it exists for just an instant, or an 11D printer if string theory is correct, or an 8==D printer if that happens to be the value in the variable D.
Skip the too-cheap entry level and get something reliable. They’re great to have handy, but easy to fall into maintenance and calibration hell. Modern 3D printers have enough sensors and smarts to self-calibrate reliably. That’s essential to make it a tool and not a tinker toy.
Same pattern in software. The best tools for specific professions usually come from someone who worked in that profession first, got frustrated with generic solutions, then built exactly what they needed.
General-purpose tools try to serve everyone and end up with features that kinda work for most cases. Someone deep in a niche builds for the edge cases they actually encounter, and it turns out those edge cases are universal within that niche - they just weren't visible to outsiders.
3d printers are great.
I attempted to productize a super niche idea I had for myself that I have ended up using multiple times. It's a simpler method of running wires without having to cut/patch drywall and drill through studs.
It started with trying to create a router-like device equipped with a motor (which was version 1 [1]).
But then iterated it to be a plate which can be attached to existing routers. That resulted in a drastically improved user experience which went from doing it all inside and making a mess to doing it in your garage with more precision. I've used the final design on 4+ projects at home and it has saved me so much time [2].
[1] https://www.dropbox.com/scl/fi/i5p330nqo0cfrfx3gxf05/Wiresha...
[2] https://trywireshark.com (includes diagrams and videos of what I 3d printed)
What is the difference between your project and regular (european) plinths, which already have a built-in channel for passing cables?
FYI: Your "Notify me when available" is landing on a 404 Not Found for me after I submit my email address
I think that is a fantastic insight that 'Making niche solutions is the point' with 3D printing.
Unfortunately, it is still very hard to _design_ niche solutions. The usability of CAD tools did not really improve at all in the last 20 years..
CAD is complicated, yes. But the biggest pain point is that engineering requires a lot of adjacent knowledge about material, tolerance, mechanical design, tooling, and so on. Making things requires patience.
CAD is the easy part.
That really depends on what you're making. I've made a lot of things that are basically, I want an object with this shape, model that shape, print that shape, success. For things where tolerances and material properties matter, a little trial and error takes care of a lot of it.
There's an engineering saying that anybody can design a bridge that won't fall down, but it takes an engineer to design a bridge that just barely won't fall down. Why do you want a bridge that just barely won't fall down? Because it's a lot cheaper to build. That's not much of a concern when you're printing little doodads at home. I waste some material by designing overly-strong structures, or getting it wrong and iterating. That's fine, the stuff is cheap.
...which is a niche problem looking for a niche solution.
My wife and I bought a Bambu P1S and I think the machine has had a duty cycle of 30% (excluding when we weren’t in town). It’s great fun. The majority of the models are ones we get from the Internet, it’s true.
But my wife used some base open source components to design a block that goes into the bits of a playpen that we used to have and transform it into something that docks with the wall instead of only with itself[0].
And I have designed with Claude a few small things like card holders for the board game power grid[1].
I wish there were better AI tools for interacting with modeling software. As it stands I use OpenSCAD with Claude and that seems as good as it can be. There are Solidworks AI startups but they’re like for professionals.
The Bambu P1S I have is quite low friction to set up. And I have an AMS2 Pro on top of it that feeds different kinds of filament (material and color) into the printer. I have just the one but now I wish I had more AMS hooked up.
0: https://wiki.roshangeorge.dev/w/Blog/2026-01-15/Modeling_Wit...
1: https://wiki.roshangeorge.dev/w/Blog/2026-01-11/Modeling_Wit...
AI coding tools are providing non-programmers giving similar ability. Not production quality, but still useful to them everyday and they can tweak as they go.
Making a good mechanical design is difficult, it usually involves making several iterations which in the physical world takes a lot more effort than iterating on software.
One thing that always bothered me is that most people with 3d printers seem to design things on their own from scratch and rarely take on others designs, improve them and share them. There is little collaboration going on for 3d prints in comparison to software. Except from maybe ~10 widely successful projects that now have healthy communities improving them.
Why is GitHub and similar sites not used more among makers?
You're assuming that CAD tooling is mature enough to enable that.
There is no standard format for CAD projects/design files. STEP is a standard format for exporting finished designs.
There are many CAD-with-code platforms, but none of them converged on a shared language the same way there are multiple competing C compilers.
I might be beating a dead horse with this one, but standardizing around FreeCAD isn't possible either, because it isn't good enough to compete with commercial CAD software like Solidworks, OnShape or Autodesk Fusion. Blender is almost there though, but for mesh-based free form 3d modeling.
Then on top of that, a 3d model on its own is useless. You need to manufacture the part. Printers have varying capabilities and something that has turned out to be particularly essential is multi color printing, since it allows embedding text and markings onto a print but not every printer is capable of doing it or doing it economically.
Ordering individual parts is expensive, which means you'd rather buy a full kit from someone who is getting volume discounts.
In your first sentence, you answered your question.
Once upon a time, the Unix philosophy was lauded in these venerated halls. "Do one thing and do it well."
Now the hype has seemed to shift to "do absolutely anything just barely well enough to get people to pay for it".
Turns out attempting to pay the bills on unix philosophy didn't go very far.
Wat? Android, Linux, BSD (incl iOS and Mac) absolutely dominate the market in terms of mind share and deployment. I can't think of another philosophy that is so directly, causally responsible for billions of dollars of realized value.
It sounds like you're parroting the corporate line of the early 80s. "Making money directly selling software artifacts is the only way to win." Which, as we know in retrospect, was a completely failed strategy, steamrolled by companies which... wait for it... adopted more flexible technology based on the Unix philosophy.
Man, that's a weird saying. Very weird.
"Unix Philosophy", which many of us wouldn't be here if it didn't exist, wasn't designed with any sort of money in mind.
That's like complaining that the company that picks up your residential trash is a shit company for not reducing your travel time to work.
Maybe not a shit company but a trash company indeed.
I'm going to lump together proprietary and for-profit software a little because I rarely see exceptions:
Proprietary / for-profit software is made that way, yes. There's no other way you can do it, given the incentive landscape. You take the huge base of FOSS code and spend as little effort as possible to build a thin little layer of moss on top. You respect the customers as little as possible to keep them around. You fix bugs as little as possible. You use other proprietary services that spy on your customers, like Sentry and Firebase, because privacy costs money.
Free / libre software has to compete on its merits. If a project isn't useful it doesn't grow. But not growing is okay. Some projects accrete over years, like a pile of stones forming a cairn. They don't need to squeeze money out of people to live because they aren't alive. They don't have to "eat".
I'm mixed on the Unix philosophy. It makes a lot of sense when you're building CLI tools that a hacker is going to plug together, because then your program is really a function in a programming environment that spans one or more entire computers. Tools like ripgrep, jq, curl, they're all great, I love them. A good function does one thing and does it well.
But just as often, I'm okay with huge software that does a lot. Web browser engines are evolving into universal GUI / IO frameworks and I'm trying to make peace with that. Systemd does a ton of stuff, but hell, so does the Linux kernel. I don't see an inherent problem with having an init system that acts like a monolith kernel for userspace. Microkernels are nifty but in the end we all ship our org charts. Maybe there's no need for microkernels if the real divide is between kernel programming and userspace programming. For the same reason, I haven't found any personal use for wasm, because wasm makes the most sense when you're connecting two pieces of code written by different teams at different times, like a GIMP plugin. I don't need wasm to plug my own code into my own code.
And for GUIs it's just been a fucking nightmare. 57 years since the Mother Of All Demos and it's still 10x easier to write `fn main()` and build a CLI program that runs on _every_ OS, rather than a GUI program that maybe runs on one OS, like it runs on Ubuntu 24.04 but not Ubuntu 22.04. What a fucking mess. GUIs don't compose, so every GUI project I've tried feels like I'm inventing the universe from scratch. It's fun but it's a stupid fucking waste of time.
Honestly browser engine frameworks like Electron and Tauri _are_ the Unix philosophy compromise. Making a GUI framework requires tens of thousands of lines of high-effort code made by experts. If a browser's one thing is "Be a GUI framework" then it allows your GUI app to just serve HTML and now it's a CLI app that runs anywhere without fucking with GTK 3 vs GTK 4 bullshit.
Sorry for the long rant. This stuff is all percolating in my head. I started with Visual Basic 6 and I still haven't seen GUIs improve since then. Phones have been a fucking step backwards, too. Everyone uses a phone but just like a mouse-and-keyboard player dominating gamepad players in a shooter game, I get a lot more done at a real desktop with a real pointing device.
I should get a 3D printer
I've been telling myself that for as long as 3D-printing has been consumer tech (about 20 years ?) and now it's shifted to "I'll borrow one my friends' printers if needs be".
In truth every time an issue fit for 3D printing has come up in my life, I solved it easily with wood and cardboard. I'm starting to recognize I might be a craftsman at heart.
I have had a similar experience; my preferred material to work with is wood. However, as I got more into tinkering with electronics and vintage computing, I'm finding more instances where wood does not achieve sufficient strength-to-weight ratio, especially for small parts where wood grain and anisotropy becomes a significant factor to consider.
Borrowing from a friend/library/work/low-cost maker space is the way to go unless you plan on printing with the thing for numerous hours per day on average. Having said that, once you start 3d printing, it becomes a tool you reach to more and more
Check your local library, a lot of them have one and can help you get started. It's usually pennies to print too. I printed an adapter for my coffee grinder at my local library a few weeks ago and it took 2 days and cost me $4. Fantastic stuff.
Even with my old/cheap Ender 3 Pro, I printed something overnight took 13.5 hrs, there it was
A side/tangent note, I wonder if it says something about myself/character, I choose the faster print than the surface finish, so I'll have this rough surface finish due to the supports. But it goes from 20hrs+ to 13.5hrs so it's like I'll take the faster option. What I printed was a shell with internal screw mounts, it was a big piece about 5.5x4x2.5" with 20% infill, regular temp/speed and the structure support for overhang.
I thought the same thing while reading this. But I worry that I'd get one and it'd just sit on a shelf somewhere collecting dust.
A whole variety of items and devices are good for this purpose, but 3D printing are an especially costly way to keep dust off your table.
You might already have one, and just don't know it :-) If you don't, it's much cheaper to get one that the author considers a 3D printer.
From TFA:
> 1. I like to think that all printers are 3D, unless it's a printer in Flatland.
It's actually a 4D printer unless it exists for just an instant, or an 11D printer if string theory is correct, or an 8==D printer if that happens to be the value in the variable D.
Skip the too-cheap entry level and get something reliable. They’re great to have handy, but easy to fall into maintenance and calibration hell. Modern 3D printers have enough sensors and smarts to self-calibrate reliably. That’s essential to make it a tool and not a tinker toy.