UPDATE - 4-Aug-03 - Red Hat Sues SCO - Put Up or Shut Up.
UPDATE - 29-Jul-03 - SCO execs are engaged in major stock maneuvers based on prices inflated by speculators.
UPDATE - 22-Jul-03 - SCO announces plan to sell UnixWare licenses to Linux users. Reaction is negative.
UPDATE - 19-Jun-03 - Identical code in Unix and Linux is apparently IBM's property.
UPDATE - 18-Jun-03 - A method is proposed by which Linux can be compared to Unix System V source code without revealing the Unix code.
UPDATE - 16-Jun-03 - SCO dclares IBM's AIX license officially revoked (IBM begs to differ) and sues Sequent, a computer company that was absorbed by IBM.
UPDATE - 13-Jun-03 - IBM Stonewalls, SCO throws tantrum.
UPDATE - 11-Jun-03 - SCO may have copied Linux code into Unix.
UPDATE - 6-Jun-03 - SCO's Dog and Pony show - two analyst sign draconian NDA
UPDATE - 4-Jun-03 - SCO's German Web site shut down buy court injunction.
UPDATE - 31-May-03 - SCO: buyout offers welcome. The truth starts to come out.
UPDATE - 29-May-03 - Lindows has a license for SCO Linux code, further confusing the issue
UPDATE A23 - 29-May-03 - SCO threatens to sue Linus Torvalds.
UPDATE - 28-May-03 - Linux Tag threatens action, and Novell claims ownership of Unix copyrights and patents.
UPDATE - 21-May-03 - SCO Shreds all evidence against Microsoft from the Caldera anti-trust case.
UPDATE - 19-May-03 - There's more! On the same day Microsoft announced licensing Unix from SCO, SCO announced compatibility with Microsoft Active Directory. Coincidence?
UPDATE - 19-May-03 - Microsoft licenses Unix code from SCO. When I wrote this article I deliberately discounted any involvement by Microsoft. Silly me.
Automation Access has sold and supported SCO Xenix and Unix continuously since 1987. While we have always considered the SCO products substandard and overpriced, software publishers refused to support any other Unix, so we support clients running SCO Open Server Unix 4.2 and 5.0 to this day.
In contrast, we have used Caldera/SCO Linux extensively since it appeared on the market and always considered it a superior product. Actions by SCO have, however caused us to abandon that product and move to SuSE Linux, a German company with a strong California presence (Caldera/SCO Linux was also developed in Germany).
SCO's recent behavior has made it impossible for us to continue supporting the company in any way. We will, however, support existing and new clients in their use of SCO products indefinitely. Clients should be aware their current SCO Unix applications should continue to function as-is until 2038 (the Unix "Y2K"). Clients on a growth path should plan migration to Linux.
SCO has embarked on irrational adventures I can only describe as suicidal, and highly damaging not only to themselves but to businesses that publish and use Unix and Linux on their systems. I don't expect SCO to still be in business three years from now (their own 10-K filing questions their viability A5), and it's possible they'll be put out of their misery a lot sooner than that.
The first questionable act to gain much notice was when SCO engaged the services of high profile lawyer David Boies, for certain "Intellectual Property matters". Boies, since his stunning victory over Microsoft in the antitrust case, has become the favorite lawyer for people with lost causes (the Al Gore campaign, for instance). This immediately raised suspicions and prejudiced many people against the company.
It turns out Boies is on contingency. Given how things are going (and
I suspect Darl McBride was not exactly up front with him either), there
may not be as much action from this source as some expected.
The first truly irrational act was to sue IBM for $1.0 Billion on vague charges of having transferred SCO intellectual property to the Linux community in violation of IBM's license for AIX Unix. SCO claims attempts to negotiate with IBM were fruitless. IBM says they were never approached at all on the matter.
If there's one thing IBM understands thoroughly, it's intellectual property - they own more of it than most entire countries, and derive considerable income from licensing it. Other things IBM understands are money and lawyers. Some speculate IBM could easily find patents or copyrights SCO incorporates without license and sue them into oblivion.
In any case, IBM has displayed no intention of rolling over and playing dead, What they seem intent on rolling over is SCO. SCO CEO Darl McBride seems quite proud that IBM intends to "blacken the Utah skies with lawyers" (A4).
IBM immediately had the case transferred from state to federal court and asked for routine extensions in their response time. The case is now expected to go to trial in about 2 years. IBM could easily stretch the trial out for 3 to 5 years. SCO will be really lucky to live that long.
The lawsuit seemed so preposterous it was immediately subjected to deep analysis by experts highly familiar with Unix and its history. Referenced here are those by open source luminaries Eric Raymond (A6), Bruce Perens (A12) and Karsten Self (A7). Agreement seems unanimous that the filing is vague, inaccurate, self serving, devoid of substantial points and that SCO will not prevail against IBM.
Detailed change logs exist so any and all additions and modifications to the Linux kernel can be tracked to date and source, so any IBM involvement can be isolated and examined by the Linux community. Outside the Linux kernel, the GNU people have kept scrupulous records and analyzed every inch of code for IP violations before issuing it.
McBride says if IBM doesn't provide them with an acceptable response by 13 June 2003, SCO will void the license to use SCO intellectual property in IBM's AIX Unix. It's just not that simple. If SCO wants to stop IBM from shipping AIX, they'll have to take the matter before a judge and request an injunction. To get the injunction, SCO will have to show concrete, verifiable evidence that IBM breached the license, and show irrevocable harm if AIX continues to ship.
SCO seems highly unlikely to sustain either point, but if SCO does not
take action in June, the world will presume they're just blowing smoke, and
life goes on. For its part, IBM says the SCO license for AIX is perpetual
and irrevocable (A16).
At the time of filing the IBM lawsuit, SCO said they were not in any way attacking the core Linux software, but just some peripheral products added by certain distributions. This changed suddenly, and SCO now claims they are finding their property shot all through the Linux kernel.
SCO recently sent a letter to 1500 companies using Linux, informing them that Linux is an illegal derivative of Unix and continued use of the offending code subjects them to liability for misappropriation of intellectual property.
As with the IBM filing, no concrete information has been provided, only vague claims of "stuff" which don't hold up to even causal analysis (A8). This is simply not the way IP infringement cases are handled. In the real world, specific examples are presented and demand is made they be licensed or removed. A negotiated settlement generally follows. Instead, SCO issues only nebulous, unsupported claims and irrational threats.
SCO said they would show the code infractions to a select group of analysts under strict supervision and an NDA (non disclosure agreement). Analysts from the big research firms and Linux organizations balked at the excessively strict NDA conditions (A30), and the fact they would have no way to know the derivation of the code (A27). A number of serious legal issues were noted, so council advised many analysts to pass up this opportunity. Two analysts eventually did view the code.
The Linux community wants specifics on code claimed in violation as soon as possible. Since all contributions to the Linux kernel are tracked, the source of anything SCO claims as their's can be identified. Most, if not all, is expected to prove legitimate, but SCO refuses to release any verifiable specifics. Meanwhile the Linux community is continuing analysis and gathering as much data as possible so they can respond quickly when they get some actual evidence (A0).
Update: - 25-May-03 - Evidence has surfaced that Novell was working on a major Linux project in 1994 while they owned Unix. If (and that's a big if) SCO actually has found some Unix code in Linux, it could have been contributed legitimately by Novell during this project. The Linux change logs should confirm this.
Why the sudden turn-around and attack on Linux? My suspicion is that, once the IBM suit was filed, SCO saw their customers and partners running for the exits, migrating to Linux on a "do it now" schedule. This is the only prudent course to take if you develop software running under SCO Unix, and evidence is now coming in to support my theory (A15).
A major problem for SCO is that the company has been distributing Linux for years, which would include the allegedly infringing code. By distributing their own version of Linux under the GPL (General Public License) they relinquish all rights to control the intellectual property contained therein. This appears to be a Catch 22 of monumental proportions.
In hope of controlling the damage, SCO announced they have withdrawn their Linux product from the market "until the intellectual property matters are clarified" - but it was still freely downloadable from their ftp site days later. It's hard to see how they can get the toothpaste back into the tube at this point, unless they think they can invalidate the GPL license.
Update 29-May-03 - The GPL licensing situation has been made even more confusing by Lindows announcing they had licensed parts of their Linux distribution from SCO (A24). Lindows says their customers are exempt from SCO enforcement actions because SCO will honor all it's contracts. This could have a serious effect on SCO's claims depending on just what and how much Lindows has licensed and distrubuted under the GPL.
SCO says withdrawal of their Linux products is only temporary, but that cannot be the case. After all this silliness there isn't anyone who'd touch SCO Linux, even if SCO were still in business. These products are totally and irrevocably dead. The future of Unix is Linux, and SCO will not be there.
Has all this caused anyone to change their mind about Linux deployment? So far, the only effect seems to be to cast doubt on SCO's viability and anger the VARs (Value Added Resellers) and developers SCO depends on for sales (A14). Many are now considering migration to Linux "pre-need".
Meanwhile, even mainstream market research houses like Meta Group and
Forester Research are advising clients to go ahead with their planned Linux
deployments (A3, A18) because
SCO is highly unlikely to prevail in court . IBM has a large vested interest
in the success of Linux, and can be expected to protect its interests to the
Why and What?
There are many conjectures floating about.
UPDATE: - 31-May-2003 - SCO's CEO Darl McBride came right out and said it - a buyout would be welcome. "If that's one of the outcomes of this, then so be it" (A26). While this certainly clarifies SCO's intentions, it may be too late. At this point, an awful lot of people want to see SCO and McBride pulverized first.
So what's going to happen? In "think like a weasel" mode, I see the value to SCO of endless FUD (Fear Uncertainty and Doubt) campaigns against Linux and cozying up to Microsoft for funding and business advantages. In contrast, there is no value in actually engaging in litigation that will certainly be lost.
Given SCO's absurd claims and nonexistent evidence, I expect the IBM suit will never go to trial. SCO will drop it with a lot of loud whining about how "against IBM's massive resources, we can't get a fair trial", thus preserving their ability to accuse IBM and Linux and issue vague threats.
Some feel they've cornered themselves into having to show real evidence come June 13th when they say they will revoke IBM's license for AIX. What I'd do (in "weasel mode") is just say, "Hey IBM, your licese is revoked and you're distributing AIX illegally". Then I'd do nothing else and let IBM figure out what to do. "After all, it's going to trial anyway".
If SCO does go to trial, IBM will crush them, and then other parties who have been injured will descend on the remains looking for retribution. Unless SCO's management actually has gone insane and has started believing their own wild claims, this will be avoided.
The best course of action for both the Linux distributions and for IBM would be to execute well planned counter suits as soon as possible. Given that SCO's claims are shaky at best and outrageous at worst, it should not be difficult to sue for interference with business, restraint of trade, defamation, misappropriation of intellectual property, or what have you.
When AT&T sued Berkeley System Design Inc. and the University of California
over AT&T code in BSD Unix, UC immediately counter sued over Berkeley code in
AT&T's System V. SCO is certainly vulnerable to similar claims, given the long
and very tangled history of Unix.
The Battle Rages On
This section is for people who would like to know more about the background leading up to the fiasco described above. It combines my personal experience with information in published timelines, particularly the one referenced as (A1).
While my outline is extremely brief and just touches on a few high points, it should give you some idea why SCO's property rights are far from clear, even had they made any attempt to state them clearly.
In 1969, Ken Thompson wrote the first version of Unix to run on the PDP-7 minicomputer at AT&T's Bell Labs.
By 1974, AT&T was licensing the Unix source code, especially to Universities, who could get it free of charge. The University of California at Berkeley soon became a major Unix developer.
In August 1980, Microsoft announced it would publish its first operating system, a version of Unix for microcomputers, to be developed mainly by The Santa Cruz Operation (SCO), a Unix porting company formed the previous year. This operating system was to be named Xenix. Xenix was to be constructed from AT&T and Berkeley code.
Just a few months later, after signing an agreement with IBM to provide an operating system called PC-DOS for the IBM PC then in development, Microsoft started to slowly lose interest in Xenix (though it continued to be used internally for software development). By 1983 Xenix was associated predominantly with SCO rather than with Microsoft.
Many versions of Unix for the PC and other microcomputers were released, but Xenix continued to be the best marketed and was ported to a number of platforms, including the Tandy (Radio Shack) PC and to Altos, which became the dominant manufacturer of multiuser computers for small business.
This widespread use of Xenix encouraged software publishers to write for it and to ignore other varieties of Unix for the PC and small business markets. Because Xenix deviated quite significantly from real Unix, this software was not easily ported to other Unix platforms, a situation SCO has always encouraged, even to the extent of subverting standards efforts such as Intel's iABI effort.
SCO went on from year to year showing losses (minimizing taxes), but when the time came for their IPO (Initial Public Offering) the books suddenly showed them quite profitable. This raised more than a few eyebrows at the time.
The IPO was successfully completed in 1993 despite the accounting questions and the resignation of CEO Larry Michels on charges of repeated and habitual sexual assault and harassment of female employees (A2).
Before the IPO, SCO would firmly correct anyone who called them "Sko" that the correct pronunciation was "Ess See Oh". Wall Street was pleased to call them "Sko", so the company adopted that pronunciation.
During this period, Automation Access was gaining experience with real Unix, starting with Microport SVr3.0. We continued to sell Xenix because many software publishers wouldn't support anything else.
PCs became more powerful so the shortcuts used by Xenix became less important. SCO licensed the Unix trademark, and in 1989 released SCO Unix System Vr3.2.x.
Automation Access recommended Esix Unix SVr3.2.x, but also sold SCO Unix r3.2.x because so many software publishers wouldn't support anything else.
I never considered SCO Unix r3.2 to be "real Unix", but rather "Xenix on steroids", and so did AT&T. They threatened to revoke SCO's license if they didn't bring it strictly up to standard. SCO stalled, because they had no intention of doing this. If they conformed, then software written for SCO Unix would also run on other brands of Unix which were better quality and less expensive.
AT&T Unix System Vr4.0 "unified Unix" was released in 1989, and included a Xenix compatibility module using code licensed from Microsoft. It also incorporated major features of Berkeley Unix and Sun Unix. PC Software publishers continued to support only SCO, even though their code would now run on real Unix.
SCO solved their license problem by promising AT&T they'd conform to standard with the release of Unix SVr4.0 - then just never upgraded to revision 4.0. They claimed their customers just didn't need it.
This left a marketing problem - everyone else was offering Unix SVr4.x. They solved this by launching a highly deceptive advertising campaign proclaiming "Version 4 Is Here!". It was actually r188.8.131.52, they just dropped the 3.2 as redundant and started calling their revisions by the third stanza, which they do to this day. Personally, I considered this maneuver to border on fraud.
Automation Access recommended Esix Unix SVr4.0.x, but also sold SCO Unix r3.2.4.x because so many software publishers wouldn't support anything else.
SCO later issued Unix v3.2.5.x with a screwy file layout and called it "Open Server". The real system files were buried 8 layers deep in a /opt directory structure, with links to them in the normal Unix places. Otherwise it was plain old SCO Unix, so why bother with the confusing mess?
In 1991, Linux Torvalds started development of Linux, based not on Unix, but on a Unix look-alike named Minix.
In 1992, Berkeley Software Design Inc. (BSDI) was formed to publish a commercial version of the University of California's Berkeley Unix. They were sued by AT&T's Unix Systems Lab. Due to technicalities, USL had to refile to include the University of California. The University of California sued USL over Berkeley features included in System V Unix and things got messy from there. The suit finally reached a negotiated settlement after the Novell purchase with removal of three files out of 18,000 and addition of a few copyright notices.
In 1993, AT&T decided it wanted out of the Unix licensing business and sold the rights to Unix to Novell. Novell was briefly in empire building mode, but was crippled by a monopoly mindset left over from their former dominance of departmental and small business networking. They renamed Unix UnixWare (to match NetWare) and allowed it to decline rapidly.
Novell's ownership caused considerable protest, so they gave the Unix trademark to the Open Group, which still has the right to say what is Unix and what is not. Sun Microsystems was particularly vocal, so Novell sold them a perpetual license for $82 Million.
In 1995, Novell, all its products in decline, bailed out and sold UnixWare to SCO, along with all the licensing rights it had purchased from AT&T. Well, no, apparently not. Apparently Novell just agreed to allow SCO to form a business around Unix licensing, while Novell still owns the patents and copyrights and receives royalties from SCO.
SCO Sued Microsoft to allow them to remove the Xenix code AT&T had put in UnixWare so they could stop paying royalties to Microsoft. SCO won.
SCO was now stuck with developing two product lines, "real Unix" and the relatively primitive SCO Unix (Open Server). They began efforts to wind down Open Server and migrate everyone to UnixWare, but once again, the software publishers refused to budge and declined to port their products to UnixWare.
Linux was now becoming a mature product, certainly rivaling SCO Open Server and barking at the heels of UnixWare. Linux included a compatibility module that allowed it to run most SCO Open Server software.
Automation Access recommended Caldera Linux, but continued to sell SCO Unix because that's what software vendors would support - but now there was a difference. The software vendors were sensing something important was going on and it was something they couldn't stop. They started becoming less and less hostile to running their software on Linux, in fact many started to openly support that usage.
SCO recognized the threat and sent a memo ridiculing Linux to their dealers, developers and customers - a memo they withdrew with apologies a few days later. Some time later I participated in an SCO conference call and Web interactive presentation for SCO dealers to explain the company's Linux strategy. It became immediately and painfully obvious they hadn't one clue in the universe what to do.
To add surrealism to the aimless and wandering talk, most of their dealers couldn't participate in the interactive part of the presentation because it required you to run a Windows PC with Internet Explorer and Microsoft Office. Now is that stupidity or what?
In those days SCO ran a road show of quarterly briefings for dealers, VARs and software developers. I attended a number of them and noticed Linux questions coming up more and more frequently. One woman said her company had ported their software to the latest version of SCO Unix, but couldn't get it to work right, apparently due to bugs in SCO Unix. She said SCO's expensive tech support provided no answers, so they tried running the product on Linux. She reported she got excellent tech support and the product was running flawlessly within hours. She told the SCO reps her company was no longer developing new software for the SCO platform.
Totally unable to formulate a strategy to counter Linux or live with it, SCO sold the entire Unix part of their business to Linux distributor Caldera (founded 1994) and renamed themselves Tarantella, after their one remaining product.
Caldera had originally been a Red Hat distributor, but started its own Linux for business distribution because it could not convince Red Hat of the need for quality and consistency.
Caldera CEO Ransom Love had a talent, a very strong talent. Unfortunately that talent was in phrasing every announcement in a way that maximized outrage among Linux enthusiasts, in marked contrast to Robert Young of Red Hat's ability to say things phrased for enthusiastic acceptance by that same group. Unfortunately for Caldera, the enthusiasts were still extremely important in introducing Linux to business.
Upon acquiring SCO, Caldera explained they would be releasing SCO's Unix intellectual properties for use in Linux, gradually, as they cleared it of possible licensing problems. The Linux enthusiasts were absolutely outraged that Caldera didn't turn all SCO Unix properties over to Linux immediately. Caldera did, however, start trickling stuff out.
Now the problem. Because Ransom Love had so screwed up Caldera's Linux marketing with his mouth, all the income was coming from the SCO products. This allowed the SCO group to take control and push the Caldera Linux people out of the company. The company newsletters started de-emphasizing the Linux products until they were barely mentioned at all.
In firm control, the SCO group renamed Caldera to SCO Group. Major problem: the SCO people still had no idea what to do with Linux, leading up to the fiasco described at the beginning of this article.
This timeline will be coming to an end soon.
- Andrew Grygus
- Automation Access
Velocity Networks: Network Consulting Service - Internet Service Provider - Web Page Design and Hosting
All trademarks and trade names are recognized as property of their respective owners