Linus Torvalds on working with NVIDIA
June 19, 2012 8:04 AM   Subscribe



 
I read this first as "Linus Torvalds is working with NVIDIA" and I was like whaaaaa
posted by DU at 8:07 AM on June 19, 2012 [1 favorite]


Photo is destined to become an Internet meme.
posted by ZenMasterThis at 8:16 AM on June 19, 2012 [1 favorite]


Wow, he's changed a lot since I saw him in Mountain View, many years ago.

He looks so, so... respectable.
posted by Malor at 8:27 AM on June 19, 2012 [4 favorites]


Linux developers have been working on independent open source driver implementations that are increasingly capable, though most are still no match for the proprietary drivers.

I'm interested to know what is limiting the progress of open source drivers.
posted by East Manitoba Regional Junior Kabaddi Champion '94 at 8:32 AM on June 19, 2012


Wow, and he is usually so mild mannered! Usually you have to beg him to tell you what he really thinks.
posted by Ad hominem at 8:33 AM on June 19, 2012 [13 favorites]


I'm interested to know what is limiting the progress of open source drivers.

Lack of access to documentation explaining how the hardware actually works is the main thing, I understand.
posted by Dysk at 8:38 AM on June 19, 2012 [4 favorites]


I doubt it's simply a "lack of documentation" since these guys can usually figure this stuff out. It's probably more like deliberate obfuscation at a all levels, from hardware, firmware, software and documentation.
posted by DU at 8:40 AM on June 19, 2012 [2 favorites]


Also, I suspect there's not a lot of "peer glamour" in developing open source drivers, as opposed to many more interesting and visible open source projects.
posted by ZenMasterThis at 8:47 AM on June 19, 2012 [1 favorite]


I doubt it's simply a "lack of documentation" since these guys can usually figure this stuff out. It's probably more like deliberate obfuscation at a all levels, from hardware, firmware, software and documentation.

Whereas if open-source driver developers had access to the documentation for these things...
posted by Dysk at 8:51 AM on June 19, 2012


You don't need to go out of your way to obfuscate complex hardware like a graphics card. That stuff is really, really obscure to start with.
posted by ryanrs at 8:51 AM on June 19, 2012 [1 favorite]


Here's the segment in which Linus expresses his displeasure with NVIDIA.
posted by Hello Dad, I'm in Jail at 8:56 AM on June 19, 2012 [1 favorite]


I give exactly zero fucks about Linux, but I can watch Torvalds talk all day. He's smart, humble and has a real sense of what he's doing.
posted by beaucoupkevin at 8:57 AM on June 19, 2012 [1 favorite]


AMD FTW! Their chips are actually slightly faster, but developers working on high-end scientific computing are more comfortable with CUDA then openCL.
Wow, he's changed a lot since I saw him in Mountain View, many years ago.

He looks so, so... respectable.
I guess the only difference is the grey hair, but yeah. That was definitely the first thing I noticed about him. I actually watched his google talk about Git, and he looked like a typical hacker, here he looks like a senator or something.

Actually he kind of looks like a good looking version of Eric Schmidt
posted by delmoi at 8:59 AM on June 19, 2012


Here's shorter clip, which includes the question.
posted by delmoi at 9:01 AM on June 19, 2012 [1 favorite]


Lack of access to documentation explaining how the hardware actually works is the main thing, I understand.

That's basically it. ATi more or less opened up its documentation, and now the open source driver is roughly on par with ATi's own driver. The open source Nvidia driver is absolutely terrible.

I doubt it's simply a "lack of documentation" since these guys can usually figure this stuff out.

What would it gain Nvidia to do that? Obfuscation is inefficient. It wastes development time and it makes both hardware and software less efficient and more prone to failure.

And ATi apparently wasn't doing that, since the open source driver improved dramatically after ATi opened up its docs. If it's such a boon to Nvidia, why wouldn't ATi have done the same?
posted by jedicus at 9:02 AM on June 19, 2012 [1 favorite]


I watched this presentation yesterday. The question he responded to references Nvidia Optimus, which is a GPU switching technology. It allows hardware manufacturers to include (more or less) 2 graphics cards and use one or the other based upon the running software, the power source etc etc.

Support on Linux has been very poor, mainly because Nvidia hid all the interfaces and only shipped horribly broken or incomplete binary blobs.

Note that Linus also mentions Android in his answer. He is pointing out that Nvidia are licencing ARM technology and coupling it with their own secret sauce (multi-core, GPU, full system on chip etc), they then package this up and sell it as an Android platform. Android works really well on ARM because the Linux kernel works really well on ARM, which it does because ARM is a really well documented platform and they've been very communicative with the kernel developers. The secret sauce Nvidia is spreading all over their ARM implementation makes it very hard to support. This is what annoys Linus.

He has always been fairly forgiving (for him) of GPU vendors being crap at drivers, mostly (I suspect) because he can ignore them. He is FAR less forgiving of people that ship system chipsets or CPUs. They're really central to the kernel and Nvidia are really bad at it.

They're probably only "the worst" because he is so exposed to them, they're in so many business segments and shipping so many units. Most ARM chips seem to come with a horrible binary firmware blob that does magic to the bundled GPU (or whatever) when you boot Linux. Its sad. Buy a raspberry pi (the great "cheap and open computer" that even Linus seems to be a broad fan of) and you're still reliant upon some binary fluff from broadcom to make it work.

Anyway, yeah... ramble and rant. I own 5 devices with Nvidia GPUs and 1 with an Nvidia CPU. So I'm only encouraging them. But given that all those devices run Linux I've had plenty of opportunity to get annoyed with them. They just live in their own little world of binary drivers and refuse to engage.

Elsewhere in the talk he mentions that Linux is really good on multi-core phones as a by-product of targeting supercomputers a decade ago. If they current "state of the art" is so hard to support and work with then the trickle down never happens. It undermines the whole Kernel model of Linux.
posted by samworm at 9:05 AM on June 19, 2012 [6 favorites]


Hmm, in the "related videos" I found this where Linus gives his thoughts on why Linux has never made much headway in the desktop market. What's interesting is that he actually says the whole "tablets replacing desktops" thing is actually good for Linux, since it's easy to get Linux to run on tablets as you don't expect them to be able to run any hardware that you could theoretically plug into a PC.

I recently had an annoying situation with Linux when I tried to install some USB wifi network adapters on some servers. I had two systems I wanted to install them on, and I decided to get two different cards in case one didn't work. They were both listed as supporting Linux.

The first one took me like a half an hour to install. You had to download a tarball off the company's website, which didn't compile. Then I went out and found drivers for the chipset which also didn't compile but somehow managed to install anyway

Then I took the second wifi card, plugged it in, and the wireless config window for Ubuntu instantly popped up.
posted by delmoi at 9:14 AM on June 19, 2012


They're probably only "the worst" because he is so exposed to them, they're in so many business segments and shipping so many units. Most ARM chips seem to come with a horrible binary firmware blob that does magic to the bundled GPU (or whatever) when you boot Linux. Its sad. Buy a raspberry pi (the great "cheap and open computer" that even Linus seems to be a broad fan of) and you're still reliant upon some binary fluff from broadcom to make it work.
How come no one ships devices like that with FPGAs? It seems like making a wifi FPGA wouldn't be that difficult, and if you change the antenna you might be able to make it work with 3g, 4g, whatever just by changing the FPGA layout.
posted by delmoi at 9:16 AM on June 19, 2012


delmoi: that's what GNU Radio / SDR (software defined radio) is all about. It seems to be experimenter-level stuff at the moment.
posted by jepler at 9:20 AM on June 19, 2012 [2 favorites]


I recently had an annoying situation with Linux when I tried to install some USB wifi network adapters on some servers. I had two systems I wanted to install them on, and I decided to get two different cards in case one didn't work. They were both listed as supporting Linux.

Are you using cards or dongles? I had a recent hair-pulling situation where I went through three dongles before I found one that could be configured easily (and not drop the connection every 10 minutes). I've got this guy working pretty well, even if it isn't getting maximum speed all the time.
posted by King Bee at 9:56 AM on June 19, 2012 [1 favorite]


I doubt it's simply a "lack of documentation" since these guys can usually figure this stuff out.

GPUs are incredibly complex, and change very quickly. Don't underestimate how hard reverse engineering is.
posted by inpHilltr8r at 9:56 AM on June 19, 2012


There's a pretty good discussion of this topic over at slashdot, including a couple of comments regarding why developing OSS drivers is not as straightforward as it appears at first glance. In addition to those two comments, I'll also point out that a modern desktop CPU has 4-8 cores, while a modern mid-range desktop GPU can have hundreds or thousands of cores.
posted by antonymous at 10:00 AM on June 19, 2012


Delmoi - many or most do contain an ARM or MIPS core, actually. (For technical reasons an FPGA-based softradio is unlikely to be competitive there.) What that means in practice, though, is just that the open-source driver writers need to write the embedded firmware that runs on the wireless adapter as well as writing the kernel driver. It allows for the occasional nifty hack but doesn't make driver writing fundamentally easier.
posted by hattifattener at 10:16 AM on June 19, 2012


When I worked at a software and OS company in Redmond, I once saw a ranked list of processes that generated a failure/crash as reported via Dr. Watson in XP.

At the top of the list—and I do mean top (like 88%)—were the various NVidia drivers.
posted by bz at 10:20 AM on June 19, 2012


Delmoi,

He also fields that question about why Linux has had trouble making inroads into the desktop market in the post video, and he says that he thinks it's because people don't want to install an OS and Linux was never bundled with hardware on a serious scale.

I think this is a big part of it, but I also thought a big part of why Linux has so far failed to conquer desktops was in an answer he gave earlier to a question about commercial use of Linux. He said that he thought it was good that companies have come in and taken up Linux because they do the "boring stuff" like QA that no one wants to do and focus on user experience that Linux developers didn't really care about.

But I think the historical dismissal of UX by Linux has been an enormous hindrance in its ability to be embraced by the average consumer. Until relatively recently Linux just was not easy to use. And there seemed to be this attitude in the Linux community that they liked the system to be hard, and if you couldn't figure it out you're an idiot so fuck off. That seems to be changing, but it's a really off-putting philosophy.
posted by Sangermaine at 10:46 AM on June 19, 2012 [2 favorites]


I'm interested to know what is limiting the progress of open source drivers.

Something like a 250:1 developer ratio for a start.

The ATI open source drivers are fairly feature complete, but their performance is way, way behind the ATI binary drivers, unless you're talking about very old hardware.
posted by pharm at 11:55 AM on June 19, 2012


I just switched from the closed-source nvidia driver to the open-source nouveau driver for nvidia cards, and I haven't found any downside. In fact, a small problem I thought was a window manager bug seems to have been fixed as a side-effect.

Linus on motivations for open source programming:
Publicly making fun of people is half the fun of open source programming.

In fact, the real reason to eschew programming in closed environments is that you can't embarrass people in public.
posted by Zed at 12:18 PM on June 19, 2012 [3 favorites]


I think this is a big part of it, but I also thought a big part of why Linux has so far failed to conquer desktops was in an answer he gave earlier to a question about commercial use of Linux. He said that he thought it was good that companies have come in and taken up Linux because they do the "boring stuff" like QA that no one wants to do and focus on user experience that Linux developers didn't really care about.

But I think the historical dismissal of UX by Linux has been an enormous hindrance in its ability to be embraced by the average consumer. Until relatively recently Linux just was not easy to use. And there seemed to be this attitude in the Linux community that they liked the system to be hard, and if you couldn't figure it out you're an idiot so fuck off. That seems to be changing, but it's a really off-putting philosophy.
Yeah, I agree. But that's starting to change with stuff like Ubuntu Unity (which hard-core types hate), if you discount Android.

There's also the issue of money, though. UI development probably isn't cheap. And it tends to be somewhat tedious
posted by delmoi at 1:45 PM on June 19, 2012 [2 favorites]


Speaking of UI development, how does ATI manage to make CCC substantially worse with virtually every single update?
posted by mek at 2:00 PM on June 19, 2012


Linus, quoted: Publicly making fun of people is half the fun of open source programming.

In fact, the real reason to eschew programming in closed environments is that you can't embarrass people in public.


...which would be a pretty simple explanation for why open source projects often do so poorly.
posted by lodurr at 2:05 PM on June 19, 2012


Ad hominem: "Wow, and he is usually so mild mannered! Usually you have to beg him to tell you what he really thinks."

I'm not sure if this is meant to be tongue-in-cheek or not. While Linus tends to be much more mild-mannered and less outspoken than many of the other "oldschool" OSS zealots, it's not difficult to gauge his opinion on any range of issues.

He's also known to offer noncommittal opinions on matters, or disagree on an issue while conceding that it's not a very important issue. I also get the impression that he doesn't really enjoy being thought of as a "leader" of the Open Source community.

However, when Linus takes a strong stand on something, it's usually a good idea to listen. He has a pretty good knack for being right about things, and carefully thinking through his opinions.

Also, the last time Linus took a really big stand on something, we ended up with Git, which is arguably starting to eclipse Linux in its influence. We should really try to piss Linus off more often, since his rage seems to translate well to innovative and high-quality software.
posted by schmod at 2:43 PM on June 19, 2012 [1 favorite]


Zed: "In fact, the real reason to eschew programming in closed environments is that you can't embarrass people in public."

Microsoft neatly circumvented this one with Windows Me!
posted by schmod at 2:43 PM on June 19, 2012 [1 favorite]


It's probably more like deliberate obfuscation at a all levels, from hardware, firmware, software and documentation.

Both unnecessary and incredibly unlikely. Staying ahead of the performance curve on commodity graphics hardware and keeping the unit cost down is already as difficult as you can possibly imagine, without having to also maintain a parallel tripwire system against reverse engineers. Time to market is everything. If the interfaces are quirky and obscure, it's going to a consequence of haste, not craftiness.
posted by George_Spiggott at 3:22 PM on June 19, 2012


schmod: Also, the last time Linus took a really big stand on something, we ended up with Git, which is arguably starting to eclipse Linux in its influence. We should really try to piss Linus off more often, since his rage seems to translate well to innovative and high-quality software.

It's important to understand, however, that this 'stand' only came about because he got absolutely screwed by the guy doing BitKeeper. Linus had been warned, and been warned, and been warned, and been warned that being dependent on a proprietary company, making himself and his team subservient to their whims, was a bad idea. He steadfastly ignored the haters, stuck with BitKeeper, and then got absolutely reamed by that company's owner when he pulled all the licenses because someone was daring to implement an open-source client for 'their' source-code management system.

Git was born out of Linus being catastrophically incorrect about the safety of depending on proprietary code.
posted by Malor at 7:23 PM on June 19, 2012 [1 favorite]


Strictly speaking, I'd say it was more a matter of being catastrophically incorrect about the safety of depending on the good will of a jackass.
posted by lodurr at 7:46 PM on June 19, 2012


But the jackass only had power over him because Linus was dependent on his proprietary code.

There's a reason why all this Free software exists, you know.
posted by Malor at 8:31 PM on June 19, 2012


Also, the last time Linus took a really big stand on something, we ended up with Git, which is arguably starting to eclipse Linux in its influence.
.... wat? The Linux kernel (which is what Linus is responsible for) powers hundreds of millions, of devices, tens of millions of cellphones, and it's practically the defacto standard for web servers. Git makes life easy for programmers, I really don't think you can compare the two at all, the Linux kernel is way more important (And in fact, the only reason git exists is to support Linus's preferred workflow for... working on the Linux kernel)
posted by delmoi at 8:46 PM on June 19, 2012 [1 favorite]


Malor: "Git was born out of Linus being catastrophically incorrect about the safety of depending on proprietary code."

There was no catastrophe. BitKeeper was the only SCM at the time that worked with the Linux kernel developers' workflow, or offered reasonable performance with Linux's massive source tree. Besides, the Linux developers had much more pressing issues to deal with than the ideological foundations of their SCM.

Once the writing was on the wall that BitKeeper would no longer be a viable option, Linus developed a better alternative in a matter of weeks.

Perhaps more impressive is the fact that Git was developed over the course of about a month. Development started on April 3, was largely completed by May, and the first Linux kernel developed with Git was released in mid-June.

Good software rarely comes out of that kind of breakneck pace, and Git is a very good piece of software.
posted by schmod at 9:16 PM on June 19, 2012


Besides, the Linux developers had much more pressing issues to deal with than the ideological foundations of their SCM.

And that bit them in the fucking ass. Not worrying about the ideological foundations of their tools put them in a very bad position, and they ended up being significantly impaired in development speed for, as I recall, at least six months, and maybe close to a year. Smaller professional development shops could die from a mistake that bad.

Linus was extremely fortunate, however in that he didn't have to pay anyone while he sorted out that mess. And it didn't take weeks, it took months... it was quite a long while before Git was fully functional and debugged.

They would have undergone tremendously less total pain if they'd either selected an existing open source alternative, or written their own tool to begin with. Their normal development could have continued normally while the tool was developed, if they decided they didn't like the premade offerings, and then they'd have had just one transition.

The only really good thing about Bitkeeper was that it taught Linus that he needed better tools.... it can be argued that Git would never have been invented, even though Linus was massively overstretched, without him first learning how useful source control is, becoming dependent on it for his workflow, and having the tool yanked out from under him.

It's funny both that one of the leading open source guys poohpoohs the ideas behind open source, and also got bitten terribly badly for making himself subservient to proprietary software. Tools that can dictate terms of use aren't true tools, they're chains.
posted by Malor at 9:59 PM on June 19, 2012 [1 favorite]


ZenMasterThis: "Also, I suspect there's not a lot of "peer glamour" in developing open source drivers, as opposed to many more interesting and visible open source projects"

One of our developers at work dabbles in X drivers. I'm not sure I buy your argument. Probably the main impediment is that there's no course undergraduates can take on the subject. Most focus on elemental algorithms and DirectX/OpenGL 3d graphics, but the underlying pipeline is poorly understood by students and faculty alike.
posted by pwnguin at 10:48 PM on June 19, 2012


Why Linux Doesn't Rule the Desktop

Case study #40813873

I've been a neck-beard for 10 years. My wife happily uses Windows 7 with little trouble and is somewhat antipathetic towards Linux.

I use the ATI OSS driver because the proprietary driver causes terrible tearing when I use VMWare. When I want to hook up my laptop to our TV, I have to switch to the proprietary ATI driver to get HDMI to work. I've written a script for this, but it requires a reboot. Then I have to fiddle with KDE4's activities for the HDMI desktop. Then I have to mess with audio device priorities.

The last time I went to hook up my laptop to the TV, my wife sweetly smiled and said, "Why don't you use my laptop? I hate to see you get so frustrated."
posted by double block and bleed at 7:28 AM on June 20, 2012


But the jackass only had power over him because Linus was dependent on his proprietary code.

No, the jackass comes first. His software could still have been f/oss and there would have been a problem. "Free" software has to be pretty fucking categorically free before the jackass ceases to come first.
posted by lodurr at 7:36 AM on June 20, 2012


Nvidia has responded, but it's pretty much PR nonsense, not a substantive response.
posted by jedicus at 7:59 AM on June 20, 2012


No, lodurr, that's the entire point of the GPL. It means people have no way to use their code as a weapon against you, as Larry McVoy did with BitKeeper.

Had it been properly GPLed, Linus would have shrugged at McVoy's tantrums, given him the "NVidia finger", and kept using it no matter what McVoy thought about it. And he could have made his own improvements and released those for everyone else to use, as well.

But, because it was proprietary, the licenses could be, and were, revoked. Suddenly, Linus wasn't allowed to use the code anymore. And his entire development process crashed in a smoking ruin, something that only large professional companies and teams running on donated time can really survive.

A little more attention to ideological foundations, and a huge amount of pain would have never happened. This stuff matters.
posted by Malor at 9:39 AM on June 20, 2012


Intel GPUs FTW! Seriously! I don't play games so who cares. The Intel drivers are fully FLOSS and quite solid. Theoretically not as fancy but the screen looks just as good when browsing web pages, coding and playing movies. NVIDIA can go take a proprietary hike.
posted by mr.ersatz at 9:58 AM on June 20, 2012


Nvidia has responded, but it's pretty much PR nonsense, not a substantive response.

I have a general rule that any verbal or written communication that uses the prepositional phrase "At the end of the day" should be viewed with extreme prejudice, because really what that should read is "At the end of the day [after doing a bunch of jackass things that make your head want to explode because we don't care about your agenda]"
posted by mcstayinskool at 10:01 AM on June 20, 2012


No, lodurr, that's the entire point of the GPL

I'm aware that's the entire point. It's a mistaken point. If a jackass controls the resources, you will still suffer, regardless of how "free" the code is.

I've pointed this out before, and I will keep doing it as long as necessary: f/oss supports large jackass players just as well as it supports small and virtuous players.

Also, your description is pretty hyperbolic. I assume that was done to enhance your point.
posted by lodurr at 10:06 AM on June 20, 2012


The Intel drivers are fully FLOSS and quite solid.

Some Intel drivers are open-source. Linux + Intel GMA 500 series video chipsets = world of pain.

If a jackass controls the resources, you will still suffer, regardless of how "free" the code is.

I'm not getting this, lodurr. If the code is free, then no one ultimately controls it and if someone wants to be a jackass, everyone else can continue to use the latest version, and fork it, and continue to develop it and move on. Contentious forks are a pain, to be sure, but there's a big difference between the pain of a contentious fork during which you can still freely use the tool, and the pain of losing access to the tool altogether.
posted by Zed at 12:26 PM on June 20, 2012


lodurr: I'm aware that's the entire point. It's a mistaken point. If a jackass controls the resources, you will still suffer, regardless of how "free" the code is.

In this world, the only unique resource is the source code. If you have access to the source code, then it cannot be taken away from you or used as a weapon against you. You have the option of continuing to use the present version forever, if you like. If you need new features or bugfixes, you can maintain it yourself, hire other people to do so, or cooperate with a group of other users to do either or both.

Typically, what happens when someone is a significant jerk, especially in a case as extreme as McVoy's, is that a bunch of users of the current tool fork the software into a new version, and add features and bugfixes over there, leaving the original tool to languish. See: X.org splitting from XFree86, as an example. The various Linux distros, by basing their windowing system on GPLed code, were able to take control of their own destinies, and switch to a fully-functional replacement for the X Window System that moved substantially faster, and served their needs much better. By making sure their infrastructure was all GPL-based, that gave (and gives) them the power to abandon teams they don't like, without having to abandon the code that team produced.

At the same time, their ability to screw over those teams is sharply limited, because if they keep using and distributing the code, those teams still benefit from any improvements that are released. So, the XFree86 team, even though they've been abandoned, can still incorporate any or all of the X.org codebase back into their original product.

Had Linus insisted on a GPL-based tool, it would have been impossible to throw a monkey-wrench into kernel development as McVoy did. And if Linus had forked a GPL-based BitKeeper, he couldn't have messed up McVoy very badly, either, because any improvements released could be incorporated into the original code base, whether Linus liked it or not.

The GPL means that jackasses will have a hell of a hard time using code to cause you grief. It renders jackassitude largely irrelevant. Whether it's corporations or individuals, their ability to force you to do anything, at least via the code you're using, is lowered to very near zero. And that's true on both sides of any distribution.... neither the writer nor the receiver can significantly damage the other via the code in question.

And that's why ideological foundations matter, and why you should take the time to think about them. Jumping into bed with proprietary software has the potential of messing up your computing life very substantially.
posted by Malor at 12:18 AM on June 21, 2012


In this world, the only unique resource is the source code. If you have access to the source code, then it cannot be taken away from you or used as a weapon against you.

Right. And if a jackass controls your repository, you don't have access.
posted by lodurr at 6:05 AM on June 21, 2012


BTW, I do think about ideological foundations. It's really presumptuous of you to assume that I don't. I also think about their weaknesses -- most obviously and notably, that they are ideological, and do not exist apart from our realizations of them. Since we are human, those realizations will be imperfect -- and indeed, they will be even more imperfect as a function of how much we rely on instrumentality to paper over human weakness.

f/oss is a great concept. But f/oss and f/oss principles do not in any way prevent jackasses from being jackasses. All that can prevent that is refusal to do business with jackasses (or taking appropriate precautions when you do -- e.g., making sure you keep archived repositories in a different location, as I'm 99.99% sure Linus and his collaborators were doing).
posted by lodurr at 6:09 AM on June 21, 2012


Right. And if a jackass controls your repository, you don't have access.

Not if you have a backup strategy, as any reasonably talented amateur, never mind a professional, would.

Further, you're deliberately conflating licensing for software with provision of storage; they're separate issues, and if you are ever in a position where a failure or deliberate jackassery by a data provider can result in permanent loss of data, then you have fouled it up. Data is easy to back up. Commercial code licenses, not so much.

BTW, I do think about ideological foundations. It's really presumptuous of you to assume that I don't.

I used that wording because you did. You said that they didn't have time to think about ideological foundations, implying that even doing so was stupid.

Well, that bit them in the ass. It bit them really badly. Had they considered their setup more carefully, McVoy would not have been able to make their lives so miserable.

Free software won't stop jackasses from being jerks, but it DOES keep them from using code as a weapon against you. Pointing out other ways they can use other services as weapons is immaterial to this observation. Free software is about code licensing.

Claiming that this doesn't also provide backups and give you redundant network services is unfair argumentation. Those are separate problems, with separate (and easy) solutions.
posted by Malor at 6:32 AM on June 21, 2012


I used that wording because you did. You said that they didn't have time to think about ideological foundations, implying that even doing so was stupid.

I'm having trouble seeing where I did that. Maybe you could help me out, here?
posted by lodurr at 7:00 AM on June 21, 2012


Claiming that this doesn't also provide backups and give you redundant network services is unfair argumentation.

It would be unfair, if I had argued that.
posted by lodurr at 7:01 AM on June 21, 2012


Weird followup story:

This was my just 7th FPP in about 8 years of being a Metafilter member.

In my life, I've received maybe 6 or 8 "cold calls" from tech recruiters I've never met offering me work.

Yesterday, one day after posting this thread, I received a LinkedIn message from a recruiter asking me if I wanted to come work for...NVIDIA.

Wat?
posted by mcstayinskool at 9:13 AM on June 21, 2012 [2 favorites]


I see how this works. There's a revolving door between MeFi posters and industry!
posted by pwnguin at 9:34 AM on June 21, 2012 [1 favorite]




« Older ♫ Welcome Back, Welcome Back, Welcome Baaaacck...   |   "On an unrelated note, this is also a great way to... Newer »


This thread has been archived and is closed to new comments