Why do proprietary drivers even exist?

[WinterMadness]

I must be lacking some kind of information. Aren't hardware manufacturers typically in the business of selling hardware? Wouldn't having open source drivers for everything they sell make more sense because it would allow the community to maintain the HW's usefulness (longevity and compatibility) thus increasing sales?

Imagine if a bug was found in a graphics card driver, and an open source coder found it and fixed it, it wouldnt take as long for them to simply investigate the given code and to implement it as opposed to trying to recreate the problem and to pinpoint it etc.

It would also allow them to sell older cards for very low prices since open source coders would probably make sure it works on modern systems.

Fill me in.

[deadflowr]

Lawyers

[1clue]

Do you remember WinModems?

There are all sorts of reasons why companies want to control the drivers for their hardware. Some of them are pretty good, some are pretty bad.

Some hardware manufacturers not only create proprietary drivers but also cooperate fully with Open Source or even make the driver themselves and open source it afterwards. Other hardware manufacturers block Open Source any way they can, for a variety of reasons.

One of those reasons is illustrated very well by the WinModem of a decade or more back. People came out with a WinModem, it was super cheap. Turns out there was barely enough hardware to make it work, and a lot was done in software. The companies who made them did not cooperate with Open Source at all, because they didn't want the world to know what a piece of crap they'd made.

Another example: IBM ported Linux to their mainframes. They Open Sourced everything except the driver for their network card, which would have revealed critical information about the card itself. Free to use (of course not valuable unless you have one of their mainframes) but not free to tinker with.

[pqwoerituytrueiwoq]

One of those reasons is illustrated very well by the WinModem of a decade or more back. People came out with a WinModem, it was super cheap. Turns out there was barely enough hardware to make it work, and a lot was done in software. The companies who made them did not cooperate with Open Source at all, because they didn't want the world to know what a piece of crap they'd made.

wouldn't this be easy to tell by dissecting one?

[Copper Bezel]

Yeah, without reading about this and based on what's stated above, that sounds a lot more like, "the sticker price reflected the development cost of software, and keeping it proprietary prevented others from selling the same product at a much lower cost," which is, you know, kinda why any proprietary software exists.

[1clue]

wouldn't this be easy to tell by dissecting one?

I suppose so, if you're a hardware type. I wouldn't have any idea looking under the covers. WinModems were half the cost of the next cheapest thing, and they were shipped with a lot of systems.

You can have whatever opinion of motives for proprietary software you want. There are some scenarios which don't seem to work well with an Open Source model, or at least which haven't yet shown any signs of working well that way. As well, programmers have every bit as much right to choose a commercial model as other programmers have to choose an Open Source model.

[bcschmerker]

Some hardware manufacturers have trade secrets to protect in the newest generation of microprocessor chips. Whereas my "Hot Rod gPC™," an Advanced Micro Devices®-equipped custom-build shoehorned into an Everex® TC2502 case, is fully FOSS-supported (xserver-xorg-video-radeon is fully functional on the integrated ATi® R600 GPU in the AMD® 780G/SB710 chipset), that's not the case with the brand-new AMD® Northern Islands GPU's in the Radeon® HD™ 7000 Series, which have massive hardware changes compared to the R6xx GPU's.

And some hardware manufacturers don't play at all nice with the open-source community; the LinUX Kernel Group had to reverse-engineer drivers for pre-2000 versions of Microsoft® Windows® NT™ for sufficient information on the registers of the nVIDIA® MCP-series chipsets in order to get said chipsets to work at all in LinUX, as I encountered during my all-too-delayed efforts to get an eMachines®/Acer® EL1210-09 (Advanced Micro Devices® Athlon 64® LE-1620, nVIDIA® MCP78S chipset; augmented with nVIDIA® GT218 in an Asus® EN210/DI/512MD3(LP) video adapter) ready for projector duty. And I still needed GeForce® Software Version 319 (ppa:ubuntu-x-swat/x-updates; nvidia-319, nvidia-settings-319) for dual-display operation, with xserver-xorg-video-nouveau unusable for either GPU (viz., no FOSS net).

[mastablasta]

also releaseing GPU drivers and hardware would give competition advantage. since thery would know how it works etc. so i kind fo understand them why they keep some thing proprietary. but i do not know why they still have drivers for old chips that are basically not sold anymore locked.

[alexan]

The "official" reason is what mastablasta said... the true reason is that if they release all specification for opensource drivers (keep in mind: no one is asking them to release the OSS of their driver... all we need it's the just the SPECS to make own drivers) they fear they will sell less. Basically because a successful hardware model can be supported unlimited with community's bugfix and other improvements.
Thus, they think, there will be lot of less people willing to buy the new version/model of the same product (mostly videographics boards).


All they want it's the "iPhone effect": want lot of people to constantly buy the +1 version of the same stuff they already own.

[ssam]

Open drivers/firmware can allow additional features that the make does not want used. If a new feature can be enabled by updating software, then people won't buy a new model. High end models may only differ in software ( http://www.argyllcms.com/doc/instruments.html#i1p2 ). Another good example is http://www.magiclantern.fm/ which adds lots of features to digital cameras.

Wireless cards are only allowed to use certain frequencies in certain countries, an open diver could use the additional frequencies or increase power levels. (actually the linux kernel has code to make sure it obeys local RF laws).