When will corporations stop applying "support" model to OSS?

Mathinker

Sun Oct 08 23:19:52 -0700 2006

manage

Right now, corporations always demand "support", even for OSS, because they lack the flexibility to adapt to the truth, which is that they don't actually need to pay for extra support for OSS in the vast majority of cases. I do not pay for support for Ubuntu, yet I get it for free in at least the same apparent quality as I get for Windows XP from Microsoft.

Judging from the high prices for OSS with support that were recently posted on Slashdot, it would be much more economical to hire consultants on a case-by-case basis whenever a support problem arises which is not taken care of through the ordinary OSS report-bug-get-fix channels (actually, I'd guess that a report-bug-post-bounty-get-fix paradigm might work even better, but OSS currently doesn't seem set up to enable it to work smoothly). IMO, (most) corporations are just not realizing yet that this is an option, because of their long experience with closed source software.

I wonder if in the very long term there will be enough redundance in OSS so that whenever a component "fails" it will be possible to just fall back on a different alternative. A scenario where there are at least two mature applications for everything you'd want to do, and the use of open standards makes it easy to move from one to the other. This utopia seems far away right now. OSS would need to have a lot (?) more manpower for it to become reality, I'd guess.

When will corporations stop applying "support" model to OSS?

Charles E Hill

Mon Oct 09 08:38:44 -0700 2006

manage

That is been my experience as well.  Who actually used support for Windows on the desktop or Office?  I mean, other than hitting MSDN or their support forums.  Microsoft's knowledgebase is an EXCELLENT resource.

I *have* used telephone support to Microsoft twice in 10 years.  One was for a post-SP hotfix that required a call to get the needed file.  It was for Win2000 Pro SP6a connecting to an AS/400 (iSeries) box.  Under certain circumstances something failed and there was a "call in to make sure you know what you are doing before we give it to you" hotfix available.

The second time was the release of NT 4.0 when purchased their "pack" of SNA, SQL, SMS, Exchange.  The SQL server disk was corrupt and I needed to call to get a replacement.

The hotfix was a free call to someone who knew what he was doing.  It was a very quick and professional experience.

The SQL disk was NOT.  This was a while ago, and they wanted to charge me $275 for the call.  I argued for 15 minutes and they would not send me the disk without a credit card.  This wouldn't happen in OSS because I'd just download another .iso.

Neither time did we have a contract.  The companies I worked for were enlightened enough to pay for a bunch of MS Press books and training classes for the staff, then let us fix it.

* * *

Now, what Linux needs is a per-incident support option.  I don't want to have to pay $1,500 per server per year to RedHat "just in case".  If I own 60 servers, that is a lot of money just for automatic updates and maybe a call.  How about $150 or $250 per incident for telephone support? (Novell charges $650). Maybe $25 - $50 per seat for automatic updates (Novell does this).

I think if that were an option, companies would sign up in droves.  Most use service and support as a "cover my ass" clause.  They do need it to be available, just in case.

a new distro with a new model

zogger

Mon Oct 09 09:53:31 -0700 2006

manage

There are lebenty leben dozen linux distros, and all of them fail it if you ask me when it comes to rewarding the devs.

What there needs to be is a distro where ALL the devs represented with the software in the release are part of a payment pool. You pay blah blah for the distro some much down to buy it and the remainder is for covering a yearly fixit cost, all the devs who sign on to be part of the distro (one CD tops) get a micropayment with every sale, maybe tabulated quarterly, in return they have to honestly work on the bugs as they show up in a timely manner and not blow people off. Man or info pages written for non coders are a must, no new features until the old bugs are fixed, and etc. that's part of the software, the dang instructions, and the current state of man pages is NOT it.. The distro issuer, the official guys putting it out, have to have a forum where the actual devs working on projects have to at least answer and respond to legit non stupid questions until such a time as their knowledge base gets built up adequately, and the search function for the knowledge base MUST WORK and be non stupid. None of this post a question and get ignored or sent off on a wild code hack case all over the intartubes-end users are not all coders! they are not all command line gurus! this is the age of GUI so the GUI has to work when you are talking "desktop". It's ridiculous to think and expect that end users are all coders or even should be if you want a desktop for the masses guy. And releases a year is WAY too much! they just get most of the bad bugs fixed and WHAM you are expected to "upgrade"-to what? The next set of betaware called a final release that ships chock fulla bugs? Who wants that beyond coding hobbiests? We already have a plethora of distros like that, there needs to be a clear alternative.

  This is arranged by the distro packager, who takes a cut for the release work and maintining the website and repositories and whatnot.. Something like the old freebooter days when the captain got a cut of such and such a size then the crew according to their rank. a working cooperative. The details would have to be worked out of course, and it wouldn't preclude the same code going into other distros eventually, but the official for-pay distro gets first dibs and the various devs coordinate their suport functions through them. Devs who wanted to work for free, no probs, their choice, devs who really wanted to get paid to code open source would get paid then.

  I would think no more than one new release a year (even that is pushing it for the end users), and maintain quality code and bug and security fixes over everything else, and take users feedback on what they did and didn't want in the distro so the next release could be arranged accordingly. If someone just coded bad stuff and abandoned it, and no one liked it, out, it gets tossed. Keeping it limited to one CD would keep the dev competition up to be included in the distro because they could get voted off the island by end user feedback if their stuff just was constantly too buggy.

I know I would pay for such a distro, I would like my dollars for sure to go to all the contributors and not just 'some" which is how it is set up now, but as it stands, there isn't a one of them out there I think is worth all that much, because there's very little difference, especially once you get into the DVD and multiple CD megabloated distros.

 And you honestly can't expect regular users to be dropping hundreds of dollars a year or even thousands, that's just absurd, it has to be affordable, perhaps a hundred a year tops, and make it on volume sales and a bounty system for any new features or applications that don't exist.

I'm sure there are additions-to and problems-with this very rough outline, just wanted to throw it out there as a possible business model for a new direction for open source. Doing it package by package is WAY too slow and lame, you can't expect end users to log into 698 different bug reporting places, or go to 698 different IRC channels and subscribe to 698 different slow weird looking email list servs.  That system is b0rken and will never work past the 1% niche market that linux on the desktop represents now, and that is one of the main reasons *why* it is b0rken.

The main benefit for end users would be true stability and quality and one stop shopping. The main benefit for devs is you actually will get paid, not like it is now with all the big major distros where only a few in house devs get paid and everyone else can go just work for free and like it. That's silly.

a new distro with a new model

Bruce Perens

Mon Oct 09 10:28:10 -0700 2006

manage

You're not the only person to think of payment pools. Many people have tried this over about 10 years. It's not worked so far.

Good developers get jobs being developers. I've employed a good many of the Debian developers through being a stockholder in Progeny Linux Systems Inc. Unfortunately, it's not made me a cent, and doesn't look like it will. But at least it's doing good.

Bruce