Taking Open Source to the Next Level

December 8, 2006 5:26 AM

Taking Open Source to the Next Level Linux? Firefox? Bah! German Markus Merz scoffs at these posers. Instead, he steps up to offer the OScar project, whose goal is to develop and build an open source *car*. While not in the same class as a Range Rover or Hummer, they hope to make something more simple and functional. This isn't the only example of hardware-based open source projects. Others include Zero Prestige, which designs kites and kite-powered vehicles, and Open Prosthetics, which offers free exchange of designs for prosthetic devices.
posted by PreacherTom (20 comments total)

Also, some hardware more in the OS software vein: OpenSparc: an open source computer processor.
posted by splatta at 5:58 AM on December 8, 2006

Open source is not magic fairy dust, and the manufacturing industry is fundamentally different from the software industry in one important way - software is essentially all design, in that a program is essentially an executable specification for what it does, whereas in manufacturing, design is "simply" the bit that preceeds the bending of metal, molding of plastic, etc, which is the unavoidably capital-intensive part of the story. Once you have written and tested a piece of software, the marginal cost of goods, ie the cost to make another copy, is practically zero. Once you have designed a car, however, the marginal cost of goods is definitely non-zero. So, what can open source do about the price of a car? It could possibly lower the design costs. Possibly.

But 10,000 geeks cannot "magic" a 200-ton hydraulic press or a spot-welding robot into existence by writing large amounts of code. Open source is a great way to create certain kinds of intellectual property, but I argue that the majority of important intellectual property that goes into the building of a car is simply not amenable to derivation via the open-source model, since most of it is gained from building physical prototypes and testing them, and furthermore that the bulk of the IP in a car (design of anti-lock brakes, design of emissions control computer, design of limited-slip differential, etc) has already been amortized across the many, many thousands of copies of each of those items, so the novel IP that goes into each new car model (car model or line, not individual car) is a relatively minor part of the cost of an (individual) car.

The one area where open source development might make a difference is in the use of "SETI@home" type computing resources crunching away at open-source finite-element models of crash tests, but that still begs the question, "how does the open source model apply to capital-intensive mass-production?"
posted by kcds at 6:30 AM on December 8, 2006 [2 favorites]

But 10,000 geeks cannot "magic" a 200-ton hydraulic press or a spot-welding robot into existence by writing large amounts of code.

But it is possible for a number of manufacturers to develop computer hardware based on the same basic specs. If a car design is open and fairly modular, can't manufacturers in various countries compete to make the best parts for it? Countries that already build big stuff can come up with the 200-ton hydraulic presses and spot-welding robots.

Maybe define a few basic classes of vehicle (light utility truck, urban family transport, urban mass transport, farm equipment, marine equipment, idiot frat boy weekend project, etc.), define the interfaces between modules, define the basic chassis, then let companies worldwide compete to make the best, cheapest components that fit the design. Other companies in various countries would assemble the parts and make sure the finished vehicles comply with local regulations. Other companies would test various builds for performance and safety.

It might not sound like good competition for Japanese manufacturers of luxury cars, but someone is going to sell gazillions of utilitarian vehicles in the developing world if they come up with a good combination of cheap, locally buildable/maintainable cars + car loan schemes the locals can afford.
posted by pracowity at 7:40 AM on December 8, 2006

Megasquirt, a do-it-yourself ECU, is a great example of an open source automotive application. I'd buy the prefab unit, but those with more time (and skillz) than money can build it themselves from components. The source code is right here. There's a big online community to help with implementation, hacks, etc. I really like the idea: Got a better algorithm for traction control? Better boost controller? Displacement on demand? Do it yourself, and share your improved code with the rest of the world.

Open source hardware -- fenders, doors, engines --gets difficult. You can open-source the design, harnessing the hive-mind to tweak the power windows or hood hinge design, maybe source a cheaper material for the fenders, etc., but like others have said, you still need a factory, presumably for-profit, at some point to make the actual parts.

It occurs to me that the automotive aftermarket already does exactly this: improves and customizes the manufacturer's design. I guess an open source car is sorta like a public aftermarket that gets involved in the original deisign. Maybe that's good enough to make it worthwhile. I can see this succeeding in China, for example.
posted by LordSludge at 8:05 AM on December 8, 2006

pracowity - car manufacturers already have common specs - NHTSA. They establish the rules about what makes cars at a minimum street legal - how wide, how well they have to handle, tolerances, safety, etc. The insurance industry picks it up from there.

The fact is that the design process for manufactured goods doesn't need to be open source, because everyone who builds cars can see the car that everyone else builds - there is no closed source.

I think what is being discussed is interchangeability among parts.

A more interesting case for OS would be an open source space exploration vehicle like a Voyager or a Mars probe, but which is standardized across all missions with common navigation systems, communications systems, image processing, etc. No "forgot to convert from inches to cm" type problems. This would drastically reduce the cost of space exploration because probes are designed almost from scratch every time. This way, if there's a problem on one mission, it can be fixed so it's never a problem on future missions.
posted by Pastabagel at 8:44 AM on December 8, 2006

Within the design period of a product is where problems get solved, and usually products are designed by a limited team of individuals. When a product is "open source" you're vastly increasing the number of potential designers, and ideas. How many times have you encountered a product where you thought you could have designed a key component better? With an open source hardware project you can take a stab at it.

Manufacturing is secondary, the design is key. This isn't really about hardware or software, it's about equal opportunity for ideas, good ones thrive, bad ones die.
posted by splatta at 8:47 AM on December 8, 2006

The prosthetics links suggest that in distorted economies like medical supplies there may well be no incentive to optimize design -- they can include an insanely expensive, proprietary rubber band not quite identical to a broccoli band. This jacks the price up for slight or no increase in functionality. So, it's not necessarily that you save money on the design process per se, but that the design process can produce a design that costs much much less to build, because the medical rubber bands aren't required. If the government or insurance companies or the military is paying for something there may well be no incentive to produce a clean design that minimizes costs.
posted by Rumple at 9:15 AM on December 8, 2006

Re: space probes, is there enough repeatability in the missions for any open source implementation to make sense? Maybe the core code could be open source -- navigation, data handling, signal transmission, that sort of thing. Sure would be cool, though, to send out, say, 50 open source probes rather than one from-scratch model.

So maybe open source could come with a timeless design, ala VW Beetle (but I'm thinkin a utilitarian 5 passenger hatchback is the place to start) -- something that will still be relevant in 10+ years -- and slowly improve it over time. Design different interchangable powertrains (gas, diesel, electric, hi-performance, high fuel economy..), braking systems, suspension designs, passive restraints, car audio setups, seating arrangements, etc..

Big project, but it sounds fun!
posted by LordSludge at 9:31 AM on December 8, 2006

I guess Markus hasn't heard of the Africar - let's wish him more success (and less fraud ...)
posted by scruss at 9:34 AM on December 8, 2006

I love the idea of open source hardware!

There are going to be issues.. Not only does hardware have the cost problems that kcds exagerates, but there are also income problems. My understanding is that the income model of open source software has been charge for support and customization, and I don't think that alone can support hardware projects.. And, yes there does have to be an income model!
Another big issue, that I won't go into right now, is safety standards.

However, there is lots of opportunity.. The cost of small run highly specialized manufacturing has plummeted with the advent CAD/CAM (simulation software, rapid prototyping, CNC machining). Also, while business has abandoned standardization at the consumer level, except when it is very marketable (USB, DVD) or completely necissary for basic functionality (NTSC, 120V AC), it has embraced standardization on another level - call it the ISO9000 version of standardization, but that doesn't quite work. There is a world of obscured standardization which allows for outsourced manufacturing and licensing of technology from small development shops. You know, globalization.

Between the reduced costs, and leveraging business to business standardization, there is room for open hardware to flourish.
posted by Chuckles at 9:59 AM on December 8, 2006

Re: space probes, is there enough repeatability in the missions for any open source implementation to make sense?

In a sense, the European Space Agency's Mars Express and Venus Express, while certainly not open source, share this concept. The Venus Express probe recycles the same design as Mars Express, and even uses some of Mars Express' spare instruments.

And, speaking of automobiles, what's the difference between a Dodge Neon and an Plymouth Neon? It's not the design, and it's not the hardware.
posted by Chinese Jet Pilot at 10:43 AM on December 8, 2006

The fact is that the design process for manufactured goods doesn't need to be open source, because everyone who builds cars can see the car that everyone else builds - there is no closed source.

Except they don't design together from the beginning so that several engine manufacturers will all, for example, get engines ready that could be swapped into the car like hard drives are installed in PCs. Instead, other manufacturers learn about the design after the fact and have to change their plans (and factory) a lot later.
posted by pracowity at 11:02 AM on December 8, 2006

I love this concept.

I've been wanting to dabble with GnuRadio for quite some time, but haven't really been able to justify the costs of getting into it.
posted by mgorsuch at 11:02 AM on December 8, 2006

There will always be costs associated with the actual manufacture of something like this, but what it could be good for is eliminating the very expensive development process. I don't have the numbers in front of me, but if memory serves, the steps involved in developing a new vehicle represent a large bulk of the cost of bringing that car to market.

An OScar would still cost money, because someone has to actually build it. But without needing to re-coup the loss of designing it, it could conceivably be sold for a fraction of what a private developer could sell it for.

Also, because it's standardized, getting new parts for it becomes much much easier, so you also reduce the cost of keeping the car functional.

I think it's a nifty idea. I hope his team pulls it off.
posted by quin at 1:06 PM on December 8, 2006

For more Open Source silicon
GOTO opencores.org.

LEON is another open SPARC v8 implementation.
posted by PROD_TPSL at 2:28 PM on December 8, 2006

When we all have rapid prototyping at home, open source car design will be huge.
posted by jiawen at 3:00 PM on December 8, 2006

An OScar would still cost money, because someone has to actually build it. But without needing to re-coup the loss of designing it, it could conceivably be sold for a fraction of what a private developer could sell it for.

Like RedHat Linux, this would be pretty cool.
posted by frogan at 3:33 PM on December 8, 2006

And, speaking of automobiles, what's the difference between a Dodge Neon and an Plymouth Neon? It's not the design, and it's not the hardware.

Branding. Marketing. (I'll grant that there may some trim level differences of which I'm unaware, as I'm not a Neon-o-phile...) I'm all for standardization, but selling essentially the same car under two different names always bugged me.
posted by LordSludge at 4:27 PM on December 8, 2006

When we all have rapid prototyping at home, open source car design will be huge.

There are people building cars in their garage RIGHT THIS SECOND. More shocking: it's been happening for decades.

So you say a water pump/alternator/engine is too complex to make/prototype at home? You can make your own aluminum foundry and shop from scrap. From there it's a matter of designing the part correctly. There's too damn many examples to list. Look around, this diy thing isn't an internet phenomenon.


But 10,000 geeks cannot "magic" a 200-ton hydraulic press or a spot-welding robot into existence by writing large amounts of code. Open source is a great way to create certain kinds of intellectual property, but I argue that the majority of important intellectual property that goes into the building of a car is simply not amenable to derivation via the open-source model, since most of it is gained from building physical prototypes and testing them,

No. It's not magic. It's hard work. Hard work a great many people are already doing. Now, not too many will get everything to build a complete car, but they won't have to. "Well, george cast the blocks, bob put together the transmissions, and joe... well he stood around drinking beer while I welded up the frames" And tell me why you couldn't use some existing parts in this application.

We already have good backyard prototyping, it's just not as rapid as you'd like, perhaps. The tools are available to anyone. Besides all this 'backyard' stuff, you've completely forgotten about the after market manufacturers who would love that slice of pie they didn't need to hand over to OEM's who own a million patents. But first, they need a market. Someone has to build enough of these things to generate interest and make no claim on the patents (which will be another, possibly insurmountable, obstacle for these guys). Eventually, entire cars would be available to everyone else. Perhaps in various stages of completion, as well.

BTW, The IP isn't as cheap as you think, hell, it's bloody expensive. It also ends up costing the consumer much more in the long run by being beholden to one OEM and its licensees for parts. The original car is just part of the story.

So, while backyard mass production isn't possible, the beginnings of something like this are. The design is the starting point. There is no car without the design.

A big obstacle will be the cost of labor here, but maybe that could be outsourced to a factory or two in mexico, except for the few diehard diy types.

Of course, it'll never happen in our lifetimes and that's too bad. Let's not forget that the automobile may soon become an even more complex creature than it already is.
posted by IronLizard at 4:30 PM on December 8, 2006

See also Open Cell Phone, which is a commodity-parts-and-open-software sorta thing. Still early on, but an interesting idea.
posted by Orb2069 at 6:30 PM on December 8, 2006