Microsoft's Response to GNU/Linux — Options
Executive summary: Microsoft will release a version of Wine and needs to buy SCO or Sun.
By Mike Banahan and Dave Fisher, 16th September, 1999.
Minor amendments 20th September 1999.
Ever since Linux started to make some ground against Microsoft, no matter how small, various attempts have been made to guess and second-guess what response the Beast of Redmond is likely to take. A lot of the analysis is written by Linux "true believers" and has been clouded by wishful thinking or a lack of experience of working in large organisations. We've tried to look outside that particular box.
The well-publicised Halloween memos did little to disabuse the Linux community of its views. Our guess is that some credence can be given to the memos and that they represent at least one strand of thinking current in the company at the time they were leaked or published, but they are assuredly not the last word.
Time has moved on and we're betting that a smarter strategy will eventually emerge from the hard-nosed regime at Microsoft.
What we've seen from Microsoft so far on this subject is unsurprising. In our view, it's been fragmentary and incoherent. You would expect this from an outfit that's desperately struggling to shed the self-inflicted millstone of the name "Windows 2000" from round its neck whilst simultaneously fighting a court battle that could threaten its existence.
The millstone is that with a name like Windows 2000, it has to ship the product within a year or become one of the most famous disasters in IT history. Anyone who has worked on a late prestige project will know the poisonous atmosphere and upper-management suck-in that it creates. Our reading is that senior management in Microsoft simply hasn't had the the time to think strategies out and set a serious response policy yet.
In frantic meetings, Linux will have been way down the agenda and given only limited amounts of thought. The easy and obvious tactics will be deployed: anything to stall and buy time will sound appealing. Hard-headed analysis by the deep thinkers will have to wait, they are too busy with really important stuff. Anyone with the intellectual horsepower to think things through will have much more important things to occupy their minds at the moment.
We think that what Microsoft has done so far is a first-rate example of a typical quick-fix: try some FUD, dismiss the "problem" in disparaging terms, try not to let it gain credence in the minds of the captive customers and hope and pray that it will just go away. The latter — hope it goes away - is usually a good tactic. Seasoned managers the world over know that this works well about ninety percent of the time. It's commonplace for this month's panic to evaporate of its own accord a few weeks later. A policy of relative inaction is often a good one.
Microsoft's difficulty is that Linux is far from going away, but gaining visibility by the day. Linux is by no means their only problem either; their attempts at domination of the handheld and PDA markets are struggling, the mobile phone manufacturers are proving very awkward, sales of desktop systems (and hence software) are flat, NT penetration has levelled off and the press seem to sense a wounded giant. The news is not good.
The company is seen by some to be structurally unstable. Most senior staff have considerable stock holdings in the company, or options on stock. Whilst the stock continues to rise this works well to lock them in, but when they sense a flat period ahead or the possibility of a downward slope to the stock price, they will cash-in as soon as they can. It is very bad news to have your senior managers suddenly develop pockets full of cash; their loyalties and interests have a habit of turning elsewhere. If they decide that it would be more interesting to get involved in a juicy start-up somewhere else, you have serious problems. Preserving its stock price is unusually important to this company.
Linux is now proving to be a stayer and is increasing its threat to Microsoft on home territory. This is not just in and of itself, but crucially because it is a form of Unix. For all the publicity you read about NT and the server market, it's worth remembering that in reality, core business systems the world over are run in one of three operating environments: IBM mainframes (and their clones), IBM AS/400s and Unix. Its image may be dull and unglamorous but corporate Unix is huge business. Suddenly through Linux Unix has been reinvented, made to look hip, trendy, streetwise, and is being given away on the cover of glossy magazines. The rate of its spread is awesome, it's more like a branch of the fashion industry than staid old mid-range operating systems. Linux has caught the imagination of legions of skilled and talented software developers, systems administrators, geeks, students and serious end-users, just at the time that Microsoft was targetting (and winning) exactly that same piece of market space.
Big corporates are starting to take note. Unix is fighting back all of a sudden: not just on technical grounds which would be easy to counter, but also at the guerilla, streetwise, zeitgeist level which is impossible to fight in traditional marketing terms. And, mostly, those who try it discover that its not just fun, but it works. Really works. Works and works and works. And where Linux doesn't deliver yet, say with scalability or non-stop-running, the traditional Unix behemoths can step in seamlessly with no need to retrain your staff, redevelop your software or disturb you users. NT's nightmare has been made flesh. Unix has turned on it suddenly and proves to have sharp teeth and claws.
The nightmare for Redmond's desktop group is only slightly further away. Linux desktops are already fully usable. The average PC user adapts in minutes (we know, we've seen it) but this market is probably safe for the short term in the developed world. The Gnome and KDE desktops are still evolving rapidly and will have to settle down to a more stable and trackable series of releases before many people will start to use them for day-to-day business applications. Until recently the same could perhaps have been said of the applications, but now the story has changed, and these applications are entirely usable even without the aid of desktops and trashcans and drag-and drop that you get with Gnome/KDE. We are seeing (and this is in the UK, which lags the States by probably two years) enough confidence amongst IT managers and directors that they are starting to think the unthinkable and plan their next generation of rollout without being dependent on Microsoft. Sun's purchase and support of Star Office has added a crucial extra layer of confidence to cement this view.
So what should Microsoft's strategy team be doing? Here is the threat in summary:
- Operating sytems on servers: severely threatened at the corporate end. Purchasers know Unix already and NT has developed a dire reputation for bloat and unreliability, let alone its security record. There is a grave risk of being marginalised. The difficulty of further product development (NT6, W2005)? is extreme; a 30 million lines-of-code tarpit has been created. Small businesses can probably still be conned into using the product but do they represent enough revenue?
- Operating systems on desktops: less immediately at risk from Linux but threatened from elsewhwere (PDAs, web appliances and the like). Incredibly rapid evolution of Linux a worry. A slow decline in sales is strongly likely unless the poor product quality and its price can be addressed. As handhelds and other devices take over, the whole concept of "desktop" could become an anachronism — perhaps within as little as five years.
- Applications: the big earner. People are wedded to their applications and would much prefer not to change. Outlook and Word addicts abound. Many large corporates have been foolish enought to make Exchange a core business service and are now well locked-in. Linux is not immediately a threat here but too many straws blow in the wind to ignore. Free office-suites are being offered (Star Office) and developed (Abi Word, GNU/KDE office, Corel etc.) to fill the vacuum left by the lack of Microsoft products. HP Open Mail supports MS Outlook and competes directly with Exchange on servers. There are almost daily product announcements of product releases like Lotus Notes, Inprise Delphi ...
Microsoft can see this just as well as you can.
The imperative for Microsoft is to preserve their brand, their revenues and their stock price to the best of their ability.
At the operating sytem end they have the choice of continuing to push the ever-increasing weight of NT uphill or getting smart. It's now reportedly some 30 million lines of code and not getting smaller. For people outside the software industry it's hard to give some idea of just what a problem that is. To our knowledge only one other software project in the commercial world has ever managed to overcome the maintenance and development problems that are caused by that kind of size (IBM OS/360) and that has been essentially frozen for years. The only way out of a pit is to stop digging. Microsoft will be forced to freeze the tarpit themselves once they realise this.
They are unlikely to turn around and adopt Linux: it would look like an admission of failure and in any case it's too far outside their control to fit their culture. They desperately need a Unix product and the fastest route is through acquisition rather than development. Buying SCO would seem the obvious choice and would, oddly, be a form of repatriation since SCO took over Microsoft's own Unix product (Xenix) years ago. A beefed-up SCO could directly target the same market as Linux and usefully create a few years of water-muddying and confusion. But buying SCO wouldn't be half as bold as going for Sun; SCO is pretty much moribund and going nowhere while Sun is a major player in a part of the market that Microsoft finds it increasingly heavy going. What's more, Sun have been goading Microsoft with Java, lawsuits and now (to cap it all) Star Office.
Sun is probably an impossible target whilst Microsoft remains a monolith — competition rules would prevent it. As several observers have already pointed out, splitting large companies often results in more overall value than keeping them together: if Microsoft has the nerve voluntarily to separate itself into an applications business and an operating system business it can probably expect to see its overall stock value increase, get the DoJ off its back and have the freedom to go after Sun or a similar juicy target.
The crown jewels are Microsoft's applications. Trying to port these to Linux/Unix would be horrendous, they are so intimately entwined with the Windows programming interfaces (APIs) nobody would ever make that attempt if there was a better way. Fortunately, there is. The Wine project already provides a Windows API wrapper, permitting native Windows application to run directly on top of Linux. It works pretty well as a demonstrator, but the project has had to reverse-engineer and second guess what really goes on behind the poorly-documented Windows API. In reality, without access to the Windows source code, it is unlikely ever to be robust enough for production use. This is of course not a problem for Microsoft: they have the source code and know what causes the problems. What's the betting that they already have a fully-working version of Wine ready and waiting for the green light for release? We think they have.
Releasing their own version of Wine too early would dangerously legitimise Linux; they are bound to wait until its invevitable. In business as in comedy, timing is everything. Once the bitter pill needs to be swallowed, we don't think they will have any compunction.