[Early April 2007: Paul Graham of Y Combinator posted Microsoft is dead, meaning they have become the established player, not a Revolutionary Force to be reckoned with.]
Following MSFT's profit growth of Jul-2005 where it dipped below 10% for the first time, I feel they are going to hit a financial 'pothole' within 2 years and a full-blown financial crisis within 5. Remember that IBM declared record profits every year until their bottom-line collapsed.
"Microsoft" as a brand will be around for a long, long time. Even if the company got bailed-out or taken over, there are too many products and too much of an installed base to evaporate quickly. If it ever became a minority platform, they would draw those wanting to be 'special and different' and enjoyed the bit-twiddling required - like Linux early adopters.
MSFT don't strike me a company that has the insight or humility to look for, or accept, someone like Lou Gerstner as IBM did. I have no personal knowledge of the MSFT corporate culture - but from the stories published, I believe that to be the single biggest obstacle to change.
Why 5 years?? Just sounded right.
I've been working in and watching the IT Industry since the early 70's.Problems with 'Fundamentals' - business models, management or technology - always catch up with companies.
And the end usually is blindingly fast - in IT change tends to be exponential.
Here are the 'Fundamentals' MSFT breaks:
- Cringely's excellent wrap on their non-investment of $60Bn accumulated earnings.
- Their arrogant culture, typified by the book title, "Everything I learned about Management, I learned from Microsoft".
- A reasonable 'cost of production' of a commodity good is 20-30% of retail price. MSFT, for it's $5Bn investment, can expect 1Bn+ sales. $5/unit in development cost. With a 50% profit-margin for them, a reasonable retail price is ~$40. If they sold direct on-line, $10/unit would give normal managers a better return.
- IT has always been about the hardware. Software is fungible - easily replaced because its invisible and intangible. Apple make their money from hardware and know what business they're in.
- 'Featurism' is no longer a consumer driver. If in 1993, Word 2.0 was underfeatured, and in 2003 Office 2003 was overblown
- when was the average optimum version of Word for users?? Around 1998 perhaps.
- Moore's Law 18month doubling of CPU speed stopped at the start of 2003. It's more like 5 years now.
[See Heb Sutter's article "Free Lunch is over"] Desktop replacement cycles have slowed considerably - and laptop sales quickened as the price and performance margin decreased.Pitching a new O/S to require new platforms is an unwelcome and unwanted strategy for users.
- In the end, computers are tools for doing other things. They are like wallpaper - everywhere and worth only occasional interest. What users *need* for any computing platform is 'it just works
- Good Software Engineering is a necessity for any large software project. Even with its much touted dumping of Longhorm and restarting Vista in Aug 2004, Jim Allchin introduced software methods that any good Open Source project would consider primitive. MSFT's "bust and tear" approach to software produces good to great user products, but defective O/S. Any software that's developed 'free to use' should never be better than any commercial offering
- Users don't care about the widgets under the hood - they need "It Just Works". Ther's a good O/S standards out there to be used - POSIX. Apple took a punt with OS/X and have really benefited.
- Layer a common O/S interface over CPU portability and the ubitquity of Intel environments/emulators and MSFT has a major exposure. If a radical new CPU design comes along, Apple will be able to support it, but MSFT won't.
- Orphaning products. Once a company drops a software product, it's made a very strong statement about what it considers the products commercial viability and usefulness - NIL. Software is critical infrastructure to its users. The source code and associated "tool chains" could/should be made public to address the 'duty of care' that major vendors have to their consumers.
what kind of "completely new processor" could that be?
postedby tom | Jan 1, 2007 | 5:16AM
'New CPU Design' - could be super-fast (e.g. incorporating GPU's), super-lowpower, massively parallel [like SIMD/Associative], loadable instruction sets[like AS/400], separate kernel & user CPU's like Seymour Cray's CDC Cyber's - or a blast from the past, like the 68000-series...
Apple is onto it's third CPU type - Microsoft has toyed with Alpha, MIPS, IA64and ??? - but not into widescale production.
I like the idea of having 3 or 4 processor types all running the same OS.
postedby steve jenkin | Jan 1, 2007 | 8:56PM