I made a fair chunk of change between 1998 and 2001, writing back-end code for companies that wanted to put some kind of service online. Trust me, the idiocy level back then was staggering. A friend and I used to say, "y'know, we could start a business that sells dollar bills for 95 cents each, and have a better business model than most of the startups out there."
The last dotcom boom was fueled by people who misunderstood the notion of branding. In practice, a brand is the reputation consumers associate with your company's name and/or logo. Apple, for instance, has a strong brand because lots of people have had good experiences with Macs and iPods, and the strength of that brand has generated lots of interest in the iPhone. Microsoft has a strong brand in the workplace, because everyone knows that MS products have been standard for the past decade or so. Both of those brands are based on people's experiences with actual products, though.
Back around 1998, marketers honestly thought the products were irrelevant. They thought the 'brand' was simply public awareness of the company name and logo. They thought you could slide products and services -- even whole business models -- in and out under that 'brand', and consumers would simply adapt to whatever was there at the moment.
They also believed in first-mover advantage.. bigtime. According to the guys in expensive suits, the first dozen or so companies who managed to establish their reputationless 'brands' would have all the time they needed to shove an actual business in under the logo. Everyone else would die because the winning 'brands' would suck up all the available customers.
So the whole dotcom bubble became a race to the peak of stupidity, and the winner (IMO) was a company that blew something like 80% of its venture funding on a single Superbowl ad that never even mentioned the company name or URL. Among the honorable (sic) mentions were boo.com and petsmart.com.
Basically, during the dotcom bubble v1.0, everyone was trying to imitate Amazon.com, which went from nothing to being one of the strongest brands in the country almost overnight. People didn't realize -- or more specifically, didn't care -- that Amazon's brand was held up by a solid reputation for good customer experiences, or that Amazon did have a business plan that led to profitability a few years down the road.
Today, everyone's trying to imitate Google, which is more or less leading the way in using the web browser as a software platform. For Google, it's just an exercise in wholesale data collection. Data allows Google to improve its search service, and the search service is a vector to sell ad impressions. Nobody else has a business model which generates profits directly from putting software-as-a-service up for use by the general public, but that doesn't stop people from trying to aggregate users with web-based software. The god news is that we have actual products these days, and that the funding is going to companies who've developed good brands based on actual user experience with the product.
The bad news is that places like twitter.com will eventually find themselves looking for enough revenue to support the millions of people who love the stuff they can get for free. It's almost inevitable that some of them will collapse. But that doesn't make this a bigger and more financially irresponsible bubble than dotcom v1.0.
I spent 1998 to 2001 walking around shaking my head at the stupidity of web ventures that could rake in tens of millions of dollars in funding, and hearing about stupider and more expensive ones every week. I'm not doing that these days. I still say, "nice service.. but I don't see how they're going to make money," every now and then, but we aren't in the middle of a balls-out silly season like we were back then.
Leave Artie MacStrawman alone. He's an idiot, and we all know it.
And as a side note: if you can only make your point by referring to Artie's behavior, please do the world a favor and just shut up. You won't be saying anything worth hearing.
So you're saying this FUD was astroturfed by Apple's PR department for the express purpose of being debunked.
Riiiiiight.
Clearly, the actual people who want to ream Apple are flippin' paragons of rational argument and carefully substantiated fact. Just look at all the "I agree with the article" posts in this thread for proof.
BTW - you might want to check your tinfoil hat.. I think the resonant internal frequency might have melted the neurons associated with critical thought.
---- If I pay for the rights to something, I've bought the right to use (and reuse) it as I will, without obstruction or degradation.
Garbage.
"Paying for the rights to use something" means buying a license, and buying a license means being bound by the terms of that license.. whatever they happen to be.
If I grant you a license to enter my property, with the restriction that you don't cut down the trees or kill any wildlife, you're bound by those terms. Period. You can't start a logging operation and fur company just by saying, "I bought the rights to use your property and refuse to accept any limitation or degradation of those rights."
And that example is precisely equivalent to buying a use license for music. The license contains terms of use, and you're bound by them, whatever they happen to be. If you don't like the terms, you can refuse to purchase the license, or you can try to negotiate a license whose terms you like better.
The courts accept a license as a public statement by the rights holder, which he can't take back, granting you freedom from prosecution or litigation as long as you obey the terms in the license. The courts also accept the act of purchasing a license as a statement that the purchaser knows the terms of the license and intends to obey them.
There's no law which says the terms have to be balanced, fair, or even sensible. I can sell you a license to listen to my music which requires you to wear a kilt and hop on one leg as long and the music is playing, and I can charge you $1000 per minute for that license. If you're fool enough to buy such a license, you're bound to those terms at that price. If you then try to tell a judge that you shouldn't be held to those terms, she'll say, "Tough. You had your chance to reject the terms before you bought the license. You did buy it, though, so get some plaid and start hopping."
The decision to sell style is one of the best business moves Apple ever made.
Selling style means you've raised the bar beyond simple functionality. Consumers appreciate an attractive package if the basic product is solid, but they resent a flaky product with go-faster stripes. They tend to feel (with justification) that you could have spent the extra money on making the damn thing work.
Apple can meet that challenge for two reasons: First, Apple sells to the high end of the market. Its margins are large enough to support the price of making everything "just work". Second, Apple controls quite a lot of its product stack, so it can make sure all the pieces fit together nicely. PC vendors have trouble selling style for exactly the same reasons. Their margins are much thinner, so the cost of making sure everything's polished will hit them where it hurts. And OEMs don't control a critical part of their product stack: the OS. It doesn't matter how good the components are or how much you've tricked out the box, a high-end Windows PC will have almost all the same issues, glitches and nuisances of a built-in-the-basement POSbox.
Apple has one more advantage, though: It has the institutional discipline to hire expert designers and then listen to what they suggest. That's very hard to do. You can be fairly sure that upper management won't start rewriting your parts specs or re-engineering the motherboard, but everybody thinks they have good taste. And the more self-deluded a company happens to be, the farther it can push patently appalling crap through the production chain before finally having to admit that nobody in their right mind would buy, say, a dog-turd brown MP3 player.
There's a difference between 'market share' and 'installed base'. 'Market share' counts the number of machines that are sold each quarter. 'Installed base' counts the number of machines already purchased and in service.
A 2-3 year lifespan is industry standard for PC hardware. Most IT departments depreciate their machines over a 3-year period, then replace them.
Macs normally see a useful life of at least 4-5 years, and that fact is reflected, among other things, in their ability to hold a good resale price. The fact that Apple continues to support hardware at least 5 years out of date in the latest versions of its software is another hint. One of the key bragging points of OS X has been that each point upgrade actually ran faster on old hardware than the previous version.
OS X 10.4.9 runs perfectly well on the 450 MHz G4 I bought back in 2000 or 2001, for instance. It doesn't transcode video or play graphics-intensive games as quickly as my new 2.1 GHz Core Duo MBP, but that's about the only performance hit I see from the older machine. User-bound tasks (word processing) and network-bound tasks (email, web browsing) fell pretty much the same either way.
The installed base of Macs is estimated to be between 10% and 15% of the market. That value follows from the sales numbers established in market share, amortized across the 5-7 year functional lifespan of the average Mac.
"One machine in ten" seems like a reasonably attractive size for a target.
Besides, you're forgetting the automated nature of malware. You don't create a botnet by hand, one machine at a time. You pump out a massive number of potential attacks and glean the ones that succeed. And having a botnet means having a massively distributed system whose resources can be devoted to making itself even bigger.
It doesn't even take an infected Mac to compromise another Mac. The attack is just a package of data, so it would be trivially easy to dedicate a Windows botnet to locating and infecting Macs if someone really wanted to.
The reason malware developers target the Windows platform is that it's so much easier to find a Windows machine with an exploitable hole and take it over. Windows up through XP carries a ton of historical baggage that assumes the existence of an isolated, single-user system: All processes are launched by a user with absolute privilege. Half the processes on any given machine are running at the highest possible level of privilege, and they accept data from sources with lower levels of privilege. The directory that contains system binaries is writable by pretty much anyone, there's no index to say where any given binary came from, and it's standard practice to add or overwrite files in that directory. The absolute-privilege daemons are controlled by the Registry, which again is writeable by almost anyone, and whose format is obscure enough that it's difficult to find tampering even if you know something is wrong with the machine.
Those were all convenient and effective solutions in the days when 99.9% of the data coming into a machine came from the person at the keyboard. But they don't fare so well against a hostile internet.
OS X doesn't have that baggage. It inherited unix's experience dealing with multi-user systems in an untrusted network environment. Yes, there are weak spots, but the attack surface is much smaller than that of Windows.
The people who collect botnets don't care about market share. They care about exploitability, especially exploitability which can be automated. Windows machines offer an easy target in that respect. Macs and unix-alike systems require more work. And there's no reason for them to do the extra work when Windows machines are both so easy to find and so easy to take over.
---- If everyone's house had no locks, they would be just as secure as if everyone's house had the best locks on the market.
I understand what you're trying to say, but there's a certain comedy value in seeing a door that's secured with a Chubb 20mm deadbolt, but framed between a pair of plate glass windows.
If we take 'security' to mean some kind of magic fairy dust you can sprinkle on part of the world to make bad things stop happening, then no.. it doesn't exist. Bruce Schneier discussed the issue at length, and quite eloquently in his book Secrets and Lies. The best approximation of 'security' we can get is a complete and integrated system whose strong points and weak points overlap each other, and whose cost/benefit ratio is proportional to the cost/risk profile of the stuff being protected.
Any such system that's tight enough to meet conventional ideas of 'security' is tough to build, and even harder to maintain. The effort and diligence curves are way above what you can expect from the everyday person on the street.
We can build systems that make it easier for people to do things that promote good security, and harder for them to do things that promote avoidable risk, but that's about the best we're ever likely to manage. Security is measured like system uptime: in orders of magnitude. One-nine security (90%) is easier to achieve than two-nines (99%), with each additional nine being harder and more expensive to tack on. It's very unlikely that we'll ever see the general public acquire the knowledge and discipline necessary to maintain overall five-nines security (99.999%), because somebody just won't think the payoff is worth the effort.
There's no comparison between that and the example given in the article.
These guys turned their site into a link farm, plain and simple. Their 'resources' page listed 329 categories of outgoing links, then they sent out spam that advertised "currently less than 50 links per category."
It's pretty tough so square that with the "victim's" mutterings that, "boy, some of the rules sure are vague."
The basic rule, "do not engage in industrial-scale stupidity" seems pretty clear to me.
---- do not hire idiot consultants to raise your pagerank.
What I want to know is: why haven't these 'victims' sued the living crap out of their 'consultant'?
I'm pretty sure the pitch session didn't run:
CONSULTANT: "For $35K, we'll set you up with a bunch of links that will drop your business right in the crapper."
CUSTOMER: "Sounds good. Here's a check."
There had to be some kind of promise that the client would get results they wanted, and that strikes me as sufficient grounds for suing the consultant to get at least the original $35K back. And given the results, it seems to me that they'd have a good shot at getting additional damages.
Of course, it's entirely possible that the contract was written with a 'no guarantee, no liability' clause, but courts chuck those out all the time. For the consultants to get away free and clear, they'd need to prove that the pitch session went more or less along the lines shown above. And if that's the case, then the client's StupidRank is right up there next to the people who lose their life savings to Nigerian 419 scams.
Gotta say, though: I wouldn't want to be either of the featured players in a Forbes article that essentialy runs, _Idiot Client Loses Tons of Money After Hiring Even Bigger Idiot as Consultant_
Well, yeah, but the principles of 'gaming' the spomsored link system are right out there in the open: Pay more, and your ad goes higher up the list. A consultant can barely spin that out long enough to warrant a coffee break.
The whole record industry has been watching album sales collapse thanks to the iTunes Store making it easy to buy single tracks, so let's bump up the bitrate on albums while leaving the price the same, and create some new demand for full-album sales.
Sounds nice in theory, but it would suck in practice.
Let's say some label pulls out a bunch of the unpublished Lena Horne tracks that have been sitting in the vaults for God knows how long, and the things end up selling like mad. Do you think Edgar Bronfman would just shrug and say, "well, we set the price at $0.25 per track, so I guess we're stuck with that."?
Not likely.
What the labels will do is track sales so they can maximize the price of anything that actually sells. And since this is all on computers, they have the capacity to adjust the numbers hour-by-hour or minute-by-minute if they really want to. It would take about ten minutes to write a script that tracks sales numbers and imposes a price bump on any song that sells more than X copies in Y amount of time.
The end result would be a system where 90% of the music online at any given time would be priced less than $0.99 per track, but 90% of the music actually sold at any given time would cost more than $0.99 per track.
And that doesn't even begin to consider the effects of per-person price hikes, "people who bought X also bought Y" hikes, etc. It would only take trivial adjustments to the software already in place to build a system guaranteed to screw you on the stuff you actually want, while still offering low, low prices on the stuff you'd never even consider buying.
And please don't bother to discuss how this would be bad for Apple. That's a given, but Apple wouldn't be the one setting prices. Prices would be set by the labels, who still think they're entitled to a $1-per-Zune fee "because we all know your customers are thieves, anyway." The labels won't give a flying damn if their pricing structure puts the iTunes Store out of business, as long as they can cream as much money as theoretically possible off its death.
You're working from the assumption that consumers value music as a physical commodity, and that just isn't the case.
In general, consumers value entertainment as a function of time. The going rate is around $5/hr.
There are just too many free variables for people to rate the value of entertainment strictly on its merits. A movie I'm willing to watch ten times might leave you cold, and vice versa. In the long run, it's easier just to assume my time is worth $N, and amortize my entertainment experience accordingly.
A movie ticket costs $7.50 where I'm from. The movie itself will average out to about 110 minutes. A DVD costs $15-20, and we can assume that people will watch a movie they buy at least twice. CDs cost $10-15 for roughly an hour's worth of music, and we can assume that most people will listen to the entire album three times, or listen to specific tracks enough to make up three hours of listening time.
$0.99 per song translates to 12 minutes of entertainment time. $1.20 translates to about 15 minutes. At an average of 3 minutes per song, people will generally feel that they're getting decent value for money on any song they'd be willing to play half a dozen times.
---- Quantum mechanics is not intuitive, but it pass every test we make with it.
Except for those ones where it has to deal with massive particles. Then it shits bricks.
Quantum Theory and General Relativity are fundamentally incompatible. QT works really well for really small and really light objects, like photons. GR works really well for really large and really heavy objects, like planets. Neither one works for really small but really heavy objects, like black holes. (I assume, but don't know offhand, that they also fail for large-but-light objects, i.e.: something with the mass of a proton and the diameter of a planet) You end up with division-by-zero errors and infinite probabilities, and all sorts of other meaningless garbage.
It's one of the embarrasments of science that both QT and GR are supported by every test we can throw at them, but we can prove mathematically that they describe fundamentally different-and-incompatible universes. We know that one or both has to be wrong, we just don't know which one, or where to look for the problem.
Sure.. just dig into any branch of QM that assumes there's only one authoritative world-line, and that non-observed states cease to exist as soon as the observation collapses the waveform.
What, exactly, makes that observation authorative, thus making observed reality fixed and determinate, which in turn keeps the past from being a multiple-choice kind of thing?
Once you've worked out the math on that, try coming up with a reason why I shouldn't hum, _You say "po-tay-to," I say "po-tah-to"_ while you explain that this is Science, not The Nature of the Word of God.
The thing that gives me headaches about observation is that it fundamentally assumes some kind of authoritative frame of reference.
To use the simplest version of Schrodinger's Cat as an example, we assume that the observer carries the authoritative frame of reference, and that the 'cat' waveform collapses to an authoritative state once the observer opens the box. The act of observation passes authority from the observer to whatever state is observed. If we make things a little more complicated with the many-worlds theory, the act of observation defines a split between two different lines of authority, one of which defines the cat as alive, the other of which defines the cat as dead.
Either way, though, something happens at the moment of observation to change the cat from an indeterminate collection of superimposed states to a fully-determined chunk of observed reality in the given authority-line. Observation is whatever happens during 'now' that separates 'the future' from 'the past'.
First of all, that's damned mysterious, and I automatically want to ask, "why?"
Second, both interpretations lead to frustrating results. If we take the assumpion that there's a single line of authoritative observation, we end up with an infinite regression of 'where did that authority come from' questions. Furthermore, assuming that there's a single authoritative frame of reference whose very existence defines reality sounds a long-winded way of saying 'God'. OTOH, if we take the many-worlds interpretation, it pretty much reduces science to a self-selecting fallacy. That which we know as 'scientific fact' is simply what we've chosen to observe, and the lack of refutation simply means that the refutation happened in some other line of authority.
Granted, I don't know enough of the math to do the really heavy lifting that might give me a better perspective on the issue. But I've also never heard any discussion of what makes observation tick. AFAIK, it's just an axiom right now.
IIRC, Apple posted revenue of something like $7B last year, while Microsoft's was around $21B. I think the profits were also around a 1:3 ratio between the two. And I know that Apple made more money last year than Dell.
Given those numbers, and considering how new Vista is, I think it's probably wrong to say that Vista has brought Microsoft more money than OS X has brought Apple, at least so far.
Apple isn't in the business of "giving you what you want." They're in the business of "selling integrated solutions."
You don't want an integrated solution? Fine. You don't want an Apple product. It's as simple as that. Rolling your own solution is great, but don't complain that you can't cherry-pick pieces from Apple's integrated product stack to play with.
Name three other categories of consumer electronics which allow you to mix and match the way you can with a computer. Nobody who buys an HDTV complains that they like the display of model X, but would really prefer to load the user interface from model Y onto it.
For that matter, name three other product categories, period. Nobody complains that they can't use BMW parts in their Camry. Nobody questions Ben & Jerry's capacity to serve the market because "I want the fruit from Cherry Garcia, but I want to put it in Hagen-Dazs chocolate chocolate chip."
The vast majority of the market wants integrated solutions, and turning everything into a pick-and-mix festival does not serve that very large chunk of the market. Most consumers don't want to be drowned in a sea of options.
Yes, vendors can push inventory onto retailers.. it's called 'flooding the channel'. Microsoft may have done that with Zunes to meet its projected sales goals, based on its sales numbers per month for the year.
There are two reasons why Apple probably hasn't done so, though:
First, you can't flood the channel continuously. About all you can do is collapse the sales you would have gotten next quarter into this month's sales report.
Say I'm a vendor shipping a product whose market demand is a million units per month, and I want to puff up my sales figures. I ship 1M in January and 1M in February, then flood the channel with 4M units in March. That lets me claim 6M sales for my first quarter. It makes a nice press release, but doesn't mean that I'll ship another 6M units next quarter. In fact, if the actual demand is only 1M per month, I won't ship any at all. I can do the same thing again in the third quarter and claim 12M in sales by September, hopefully snagging some consumer interest for the upcoming Christmas season. But by the end of the year, I'll still only have shipped 12M units.
Bottom line, though: nobody can flood a channel with even a significant fraction of 100M units if the product doesn't already sell damned well.
The second problem is inventory management: Apple gives Dell a run for its money in keeping the supply chain thin. Apple itself owns less than 2 days worth of inventory at any given time, IIRC.
But in the example I gave above, I needed 6M units by the end of March. If I also want to stay below 2 days of unshipped inventory, I'd need a factory that can produce 1M units/month for two months, then jump to 4M units/month for March. Then I'd have to shut the whole thing down from April through June and start the whole thing up again in July.
It's incredibly difficult and expensive to jerk a factory's production capacity around like that. You can't build the product without machinery, which is expensive and can't just be rented for a month (who's going to have that much equipment stiiting around idle?). You need labor, which requires training (another sunk expense of both money and time) and doesn't like to be laid off every other quarter. And you need components, which means your suppliers would have to be willing to deal with exactly the same problems themselves.
Actually, the theory I've heard is that SCO wants to get its hands on the private (protected) documents belonging to IBM's lawyers.
Normally, communication between a lawyer and their client is privileged. The other side can't see it and can't subpoena it, no matter how much they want to or how much it might help their case. You can admit being guilty as sin to your lawyer, and that admission will never be allowed to enter a court of law.
Thing is, the privilege only exists if that information stays strictly between you and your lawyer. If anyone else gets their hands on information that was supposed to be protected by client-attorney privilege, the court will assume that you've waived the privilege and are willing to have the information appear in court.
If SCO could show that IBM or its lawyers passed privileged information to Groklaw, SCO could ask the court for permission to see all of IBM's privileged documents.
That would essentially take the case back to day zero. There would be a whole new round of motions to define the scope of privileged documents SCO was allowed to see, more discovery, more depositions, another chance for SCO to redefine its claims against IBM, and possibly (a very remote chance) give SCO some evidence that IBM actually did do something arguably illegal somewhere along the way.
At minimum, it would mean tons more delay. And SCO would also get an excuse to unleash all manner of punitive crap on PJ for investing the time, effort, and skill necessary to blow their media-driven FUD campaign out of the water.
It's important to remember that SCO's FUD was working until Groklaw came along. Journalists were still writing "there's lots of smoke, so there must be some kind of fire," articles even a couple of years after Groklaw got started. It was only when PJ piled up a big enough heap of documents, cross-referenced the claims, and then started collecting evidence to show that SCO might not be telling the whole truth, that the media machine stopped dancing to SCO's tune. It wasn't until journalists started actually looking at the evidence and drawing their own conclusions, instead of just parroting SCO's latest set of talking points or paraphrasing whatever article SCO had astroturfed out through its shills, that SCO really started to lose its case in the court of public opinion.
3. Transfer my efforts to a line of work society values more highly.
That's how economies grow.
Back in the 1950s, companies had whole rooms full of people punching adding machines. In the 1900s, it was all done in longhand. Today, a single person with a spreadsheet can do more math in 10 seconds than all the clerks in the world could do a hundred years ago. A six-man construction crew with backhoes and dump trucks can do more work in a week than a hundred ditch-diggers and dirt-carriers could do in a year.
The people who were displaced from those careers didn't just curl up and die, though. They found other jobs. And since we don't have to pay a few million people just to keep the books in order these days, companies can sell their goods and services at lower prices and still make a profit. Consumers get better quality goods for lower prices, meaning the dollar they earned doing something more valuable than digging a ditch goes farther than it ever did before.
In fact, we can engage in types of business that would never be considered under the old system. Can you imagine trying to run something like the iTunes store or Amazon with a bazillion human clerks crunching the numbers and keeping the inventory straight?
I think you'll find that the labels are the ones that have set the regional limits and pricing standards. Apple is bound by the contracts the labels were willing to negotiate, and the labels didn't want to negotiate liberal contracts when the iTunes stores were first being set up.
Having to run multiple, mutually exclusive stores is probably a dead loss for Apple all around. There's the massive duplication of effort in making each store run and managing the inventories, there's the effort of barring people in one region from using the store for another region, and there's the dissatisfaction from customers who can't get the music they want if it's only for sale in another region.
Apple runs the iTunes store as a value-added service for the iPod. The more music that's available, and the easier the stuff is to obtain, the more value it adds. How could it possibly hurt Apple to run a single store for everyone in the world, with all the music equally availble to everyone?
Given the track records of the players in question, I doubt that an investigation will find that Apple were the ones who went to the negotiating table saying, "hey, let's waste a lot of resources and piss off a lot of customers by making a patchwork of regional stores, offering different inventories at different prices in each one, and making people in one region wait six months longer to get access to their store than their neighbors 50 kilometers away!"
Actually, Apple is a vertically integrated systems vendor. The hardware's where they make the money, and the software is what makes the hardware worth buying.
Meanwhile, Microsoft is more like a component vendor to OEMs than a software house per se. Most people don't buy Windows, they get it bundled when they buy a computer. Most Office licenses (Microsoft's only other product that generates a profit) are N-hundred-seat deals, paid for by companies whose IT departments lay their own bundled software installation over the OEM's bundle before giving the machines to the workers.
Microsoft's revenue for 2006 was something like $21 billion, most of which probably came from less than a thousand major clients. Apple's revenue for 2006 was something like $7 billion, most of which was from people buying individual devices. Neither business model is inherently 'better' than the other, they're just different. Each is profitable in its own way, and each has its strengths and weaknesses.
The place where Apple is really hurting Microsoft is indirectly, in the midrange hardware market.
OEMs like Dell and HP have huge sales volumes because they offer a tier of low priced machines. That $300 PC isn't bringing Dell enough money to make running the production line worthwhile, but it does draw customers to the brand, and it allows Dell to buy enough parts to get good economies of scale. Then Dell uses those same (cheap-in-volume) parts for its machines in the $1500-5000 range, and that's where Dell actually makes its money.
But now Apple is competing hard in the $1500-5000 price range, and every sale Dell loses to the Mac screws with Dell's ability to keep the $300 machine lines running. But without the those cheap machines, Dell can't get the volume discounts, which makes it that much harder for Dell to keep things going. Basically, when you've built your business on leveraging economies of scale, losing those economies of scale is a bitch.
Thing is, Microsoft doesn't care about the machine's unit price. It just sells Windows as a component, so Microsoft makes $N per machine whether that machine sells for $300 or $3000. In other words, a 1-2% loss in the midrange market hurts Dell a whole lot more than it hurts Microsoft, but dumping the cheapest 25% of Dell's total market share would hurt Microsoft a whole lot more than it would hurt Dell.
Back when Microsoft was trying to kill Netscape, Microsoft execs used the term, "cutting off their oxygen." In the long run, midrange hardware sales are the oxygen for OEM's like Dell, and OEMs like Dell are Microsoft's oxygen.
Slashdot Mirror
Amen to that..
I made a fair chunk of change between 1998 and 2001, writing back-end code for companies that wanted to put some kind of service online. Trust me, the idiocy level back then was staggering. A friend and I used to say, "y'know, we could start a business that sells dollar bills for 95 cents each, and have a better business model than most of the startups out there."
The last dotcom boom was fueled by people who misunderstood the notion of branding. In practice, a brand is the reputation consumers associate with your company's name and/or logo. Apple, for instance, has a strong brand because lots of people have had good experiences with Macs and iPods, and the strength of that brand has generated lots of interest in the iPhone. Microsoft has a strong brand in the workplace, because everyone knows that MS products have been standard for the past decade or so. Both of those brands are based on people's experiences with actual products, though.
Back around 1998, marketers honestly thought the products were irrelevant. They thought the 'brand' was simply public awareness of the company name and logo. They thought you could slide products and services -- even whole business models -- in and out under that 'brand', and consumers would simply adapt to whatever was there at the moment.
They also believed in first-mover advantage.. bigtime. According to the guys in expensive suits, the first dozen or so companies who managed to establish their reputationless 'brands' would have all the time they needed to shove an actual business in under the logo. Everyone else would die because the winning 'brands' would suck up all the available customers.
So the whole dotcom bubble became a race to the peak of stupidity, and the winner (IMO) was a company that blew something like 80% of its venture funding on a single Superbowl ad that never even mentioned the company name or URL. Among the honorable (sic) mentions were boo.com and petsmart.com.
Basically, during the dotcom bubble v1.0, everyone was trying to imitate Amazon.com, which went from nothing to being one of the strongest brands in the country almost overnight. People didn't realize -- or more specifically, didn't care -- that Amazon's brand was held up by a solid reputation for good customer experiences, or that Amazon did have a business plan that led to profitability a few years down the road.
Today, everyone's trying to imitate Google, which is more or less leading the way in using the web browser as a software platform. For Google, it's just an exercise in wholesale data collection. Data allows Google to improve its search service, and the search service is a vector to sell ad impressions. Nobody else has a business model which generates profits directly from putting software-as-a-service up for use by the general public, but that doesn't stop people from trying to aggregate users with web-based software. The god news is that we have actual products these days, and that the funding is going to companies who've developed good brands based on actual user experience with the product.
The bad news is that places like twitter.com will eventually find themselves looking for enough revenue to support the millions of people who love the stuff they can get for free. It's almost inevitable that some of them will collapse. But that doesn't make this a bigger and more financially irresponsible bubble than dotcom v1.0.
I spent 1998 to 2001 walking around shaking my head at the stupidity of web ventures that could rake in tens of millions of dollars in funding, and hearing about stupider and more expensive ones every week. I'm not doing that these days. I still say, "nice service.. but I don't see how they're going to make money," every now and then, but we aren't in the middle of a balls-out silly season like we were back then.
Leave Artie MacStrawman alone. He's an idiot, and we all know it.
And as a side note: if you can only make your point by referring to Artie's behavior, please do the world a favor and just shut up. You won't be saying anything worth hearing.
So you're saying this FUD was astroturfed by Apple's PR department for the express purpose of being debunked.
Riiiiiight.
Clearly, the actual people who want to ream Apple are flippin' paragons of rational argument and carefully substantiated fact. Just look at all the "I agree with the article" posts in this thread for proof.
BTW - you might want to check your tinfoil hat.. I think the resonant internal frequency might have melted the neurons associated with critical thought.
---- If I pay for the rights to something, I've bought the right to use (and reuse) it as I will, without obstruction or degradation.
Garbage.
"Paying for the rights to use something" means buying a license, and buying a license means being bound by the terms of that license.. whatever they happen to be.
If I grant you a license to enter my property, with the restriction that you don't cut down the trees or kill any wildlife, you're bound by those terms. Period. You can't start a logging operation and fur company just by saying, "I bought the rights to use your property and refuse to accept any limitation or degradation of those rights."
And that example is precisely equivalent to buying a use license for music. The license contains terms of use, and you're bound by them, whatever they happen to be. If you don't like the terms, you can refuse to purchase the license, or you can try to negotiate a license whose terms you like better.
The courts accept a license as a public statement by the rights holder, which he can't take back, granting you freedom from prosecution or litigation as long as you obey the terms in the license. The courts also accept the act of purchasing a license as a statement that the purchaser knows the terms of the license and intends to obey them.
There's no law which says the terms have to be balanced, fair, or even sensible. I can sell you a license to listen to my music which requires you to wear a kilt and hop on one leg as long and the music is playing, and I can charge you $1000 per minute for that license. If you're fool enough to buy such a license, you're bound to those terms at that price. If you then try to tell a judge that you shouldn't be held to those terms, she'll say, "Tough. You had your chance to reject the terms before you bought the license. You did buy it, though, so get some plaid and start hopping."
No, you're not; and yes, there is. Especially when the tease contains high-test flamebait like, "They just couldn't compete under all the iPod hype."
Of course troll-whoring like that is pretty much guaranteed to give Slashdot's editors a stiffy, so of course it got approved.
The decision to sell style is one of the best business moves Apple ever made.
Selling style means you've raised the bar beyond simple functionality. Consumers appreciate an attractive package if the basic product is solid, but they resent a flaky product with go-faster stripes. They tend to feel (with justification) that you could have spent the extra money on making the damn thing work.
Apple can meet that challenge for two reasons: First, Apple sells to the high end of the market. Its margins are large enough to support the price of making everything "just work". Second, Apple controls quite a lot of its product stack, so it can make sure all the pieces fit together nicely. PC vendors have trouble selling style for exactly the same reasons. Their margins are much thinner, so the cost of making sure everything's polished will hit them where it hurts. And OEMs don't control a critical part of their product stack: the OS. It doesn't matter how good the components are or how much you've tricked out the box, a high-end Windows PC will have almost all the same issues, glitches and nuisances of a built-in-the-basement POSbox.
Apple has one more advantage, though: It has the institutional discipline to hire expert designers and then listen to what they suggest. That's very hard to do. You can be fairly sure that upper management won't start rewriting your parts specs or re-engineering the motherboard, but everybody thinks they have good taste. And the more self-deluded a company happens to be, the farther it can push patently appalling crap through the production chain before finally having to admit that nobody in their right mind would buy, say, a dog-turd brown MP3 player.
There's a difference between 'market share' and 'installed base'. 'Market share' counts the number of machines that are sold each quarter. 'Installed base' counts the number of machines already purchased and in service.
A 2-3 year lifespan is industry standard for PC hardware. Most IT departments depreciate their machines over a 3-year period, then replace them.
Macs normally see a useful life of at least 4-5 years, and that fact is reflected, among other things, in their ability to hold a good resale price. The fact that Apple continues to support hardware at least 5 years out of date in the latest versions of its software is another hint. One of the key bragging points of OS X has been that each point upgrade actually ran faster on old hardware than the previous version.
OS X 10.4.9 runs perfectly well on the 450 MHz G4 I bought back in 2000 or 2001, for instance. It doesn't transcode video or play graphics-intensive games as quickly as my new 2.1 GHz Core Duo MBP, but that's about the only performance hit I see from the older machine. User-bound tasks (word processing) and network-bound tasks (email, web browsing) fell pretty much the same either way.
Define 'nominal'.
The installed base of Macs is estimated to be between 10% and 15% of the market. That value follows from the sales numbers established in market share, amortized across the 5-7 year functional lifespan of the average Mac.
"One machine in ten" seems like a reasonably attractive size for a target.
Besides, you're forgetting the automated nature of malware. You don't create a botnet by hand, one machine at a time. You pump out a massive number of potential attacks and glean the ones that succeed. And having a botnet means having a massively distributed system whose resources can be devoted to making itself even bigger.
It doesn't even take an infected Mac to compromise another Mac. The attack is just a package of data, so it would be trivially easy to dedicate a Windows botnet to locating and infecting Macs if someone really wanted to.
The reason malware developers target the Windows platform is that it's so much easier to find a Windows machine with an exploitable hole and take it over. Windows up through XP carries a ton of historical baggage that assumes the existence of an isolated, single-user system: All processes are launched by a user with absolute privilege. Half the processes on any given machine are running at the highest possible level of privilege, and they accept data from sources with lower levels of privilege. The directory that contains system binaries is writable by pretty much anyone, there's no index to say where any given binary came from, and it's standard practice to add or overwrite files in that directory. The absolute-privilege daemons are controlled by the Registry, which again is writeable by almost anyone, and whose format is obscure enough that it's difficult to find tampering even if you know something is wrong with the machine.
Those were all convenient and effective solutions in the days when 99.9% of the data coming into a machine came from the person at the keyboard. But they don't fare so well against a hostile internet.
OS X doesn't have that baggage. It inherited unix's experience dealing with multi-user systems in an untrusted network environment. Yes, there are weak spots, but the attack surface is much smaller than that of Windows.
The people who collect botnets don't care about market share. They care about exploitability, especially exploitability which can be automated. Windows machines offer an easy target in that respect. Macs and unix-alike systems require more work. And there's no reason for them to do the extra work when Windows machines are both so easy to find and so easy to take over.
---- If everyone's house had no locks, they would be just as secure as if everyone's house had the best locks on the market.
I understand what you're trying to say, but there's a certain comedy value in seeing a door that's secured with a Chubb 20mm deadbolt, but framed between a pair of plate glass windows.
If we take 'security' to mean some kind of magic fairy dust you can sprinkle on part of the world to make bad things stop happening, then no.. it doesn't exist. Bruce Schneier discussed the issue at length, and quite eloquently in his book Secrets and Lies. The best approximation of 'security' we can get is a complete and integrated system whose strong points and weak points overlap each other, and whose cost/benefit ratio is proportional to the cost/risk profile of the stuff being protected.
Any such system that's tight enough to meet conventional ideas of 'security' is tough to build, and even harder to maintain. The effort and diligence curves are way above what you can expect from the everyday person on the street.
We can build systems that make it easier for people to do things that promote good security, and harder for them to do things that promote avoidable risk, but that's about the best we're ever likely to manage. Security is measured like system uptime: in orders of magnitude. One-nine security (90%) is easier to achieve than two-nines (99%), with each additional nine being harder and more expensive to tack on. It's very unlikely that we'll ever see the general public acquire the knowledge and discipline necessary to maintain overall five-nines security (99.999%), because somebody just won't think the payoff is worth the effort.
There's no comparison between that and the example given in the article.
These guys turned their site into a link farm, plain and simple. Their 'resources' page listed 329 categories of outgoing links, then they sent out spam that advertised "currently less than 50 links per category."
It's pretty tough so square that with the "victim's" mutterings that, "boy, some of the rules sure are vague."
The basic rule, "do not engage in industrial-scale stupidity" seems pretty clear to me.
---- do not hire idiot consultants to raise your pagerank.
What I want to know is: why haven't these 'victims' sued the living crap out of their 'consultant'?
I'm pretty sure the pitch session didn't run:
CONSULTANT: "For $35K, we'll set you up with a bunch of links that will drop your business right in the crapper."
CUSTOMER: "Sounds good. Here's a check."
There had to be some kind of promise that the client would get results they wanted, and that strikes me as sufficient grounds for suing the consultant to get at least the original $35K back. And given the results, it seems to me that they'd have a good shot at getting additional damages.
Of course, it's entirely possible that the contract was written with a 'no guarantee, no liability' clause, but courts chuck those out all the time. For the consultants to get away free and clear, they'd need to prove that the pitch session went more or less along the lines shown above. And if that's the case, then the client's StupidRank is right up there next to the people who lose their life savings to Nigerian 419 scams.
Gotta say, though: I wouldn't want to be either of the featured players in a Forbes article that essentialy runs, _Idiot Client Loses Tons of Money After Hiring Even Bigger Idiot as Consultant_
Well, yeah, but the principles of 'gaming' the spomsored link system are right out there in the open: Pay more, and your ad goes higher up the list. A consultant can barely spin that out long enough to warrant a coffee break.
addendum:
The whole record industry has been watching album sales collapse thanks to the iTunes Store making it easy to buy single tracks, so let's bump up the bitrate on albums while leaving the price the same, and create some new demand for full-album sales.
Sounds nice in theory, but it would suck in practice.
Let's say some label pulls out a bunch of the unpublished Lena Horne tracks that have been sitting in the vaults for God knows how long, and the things end up selling like mad. Do you think Edgar Bronfman would just shrug and say, "well, we set the price at $0.25 per track, so I guess we're stuck with that."?
Not likely.
What the labels will do is track sales so they can maximize the price of anything that actually sells. And since this is all on computers, they have the capacity to adjust the numbers hour-by-hour or minute-by-minute if they really want to. It would take about ten minutes to write a script that tracks sales numbers and imposes a price bump on any song that sells more than X copies in Y amount of time.
The end result would be a system where 90% of the music online at any given time would be priced less than $0.99 per track, but 90% of the music actually sold at any given time would cost more than $0.99 per track.
And that doesn't even begin to consider the effects of per-person price hikes, "people who bought X also bought Y" hikes, etc. It would only take trivial adjustments to the software already in place to build a system guaranteed to screw you on the stuff you actually want, while still offering low, low prices on the stuff you'd never even consider buying.
And please don't bother to discuss how this would be bad for Apple. That's a given, but Apple wouldn't be the one setting prices. Prices would be set by the labels, who still think they're entitled to a $1-per-Zune fee "because we all know your customers are thieves, anyway." The labels won't give a flying damn if their pricing structure puts the iTunes Store out of business, as long as they can cream as much money as theoretically possible off its death.
You're working from the assumption that consumers value music as a physical commodity, and that just isn't the case.
In general, consumers value entertainment as a function of time. The going rate is around $5/hr.
There are just too many free variables for people to rate the value of entertainment strictly on its merits. A movie I'm willing to watch ten times might leave you cold, and vice versa. In the long run, it's easier just to assume my time is worth $N, and amortize my entertainment experience accordingly.
A movie ticket costs $7.50 where I'm from. The movie itself will average out to about 110 minutes. A DVD costs $15-20, and we can assume that people will watch a movie they buy at least twice. CDs cost $10-15 for roughly an hour's worth of music, and we can assume that most people will listen to the entire album three times, or listen to specific tracks enough to make up three hours of listening time.
$0.99 per song translates to 12 minutes of entertainment time. $1.20 translates to about 15 minutes. At an average of 3 minutes per song, people will generally feel that they're getting decent value for money on any song they'd be willing to play half a dozen times.
---- Quantum mechanics is not intuitive, but it pass every test we make with it.
Except for those ones where it has to deal with massive particles. Then it shits bricks.
Quantum Theory and General Relativity are fundamentally incompatible. QT works really well for really small and really light objects, like photons. GR works really well for really large and really heavy objects, like planets. Neither one works for really small but really heavy objects, like black holes. (I assume, but don't know offhand, that they also fail for large-but-light objects, i.e.: something with the mass of a proton and the diameter of a planet) You end up with division-by-zero errors and infinite probabilities, and all sorts of other meaningless garbage.
It's one of the embarrasments of science that both QT and GR are supported by every test we can throw at them, but we can prove mathematically that they describe fundamentally different-and-incompatible universes. We know that one or both has to be wrong, we just don't know which one, or where to look for the problem.
Sure.. just dig into any branch of QM that assumes there's only one authoritative world-line, and that non-observed states cease to exist as soon as the observation collapses the waveform.
What, exactly, makes that observation authorative, thus making observed reality fixed and determinate, which in turn keeps the past from being a multiple-choice kind of thing?
Once you've worked out the math on that, try coming up with a reason why I shouldn't hum, _You say "po-tay-to," I say "po-tah-to"_ while you explain that this is Science, not The Nature of the Word of God.
The thing that gives me headaches about observation is that it fundamentally assumes some kind of authoritative frame of reference.
To use the simplest version of Schrodinger's Cat as an example, we assume that the observer carries the authoritative frame of reference, and that the 'cat' waveform collapses to an authoritative state once the observer opens the box. The act of observation passes authority from the observer to whatever state is observed. If we make things a little more complicated with the many-worlds theory, the act of observation defines a split between two different lines of authority, one of which defines the cat as alive, the other of which defines the cat as dead.
Either way, though, something happens at the moment of observation to change the cat from an indeterminate collection of superimposed states to a fully-determined chunk of observed reality in the given authority-line. Observation is whatever happens during 'now' that separates 'the future' from 'the past'.
First of all, that's damned mysterious, and I automatically want to ask, "why?"
Second, both interpretations lead to frustrating results. If we take the assumpion that there's a single line of authoritative observation, we end up with an infinite regression of 'where did that authority come from' questions. Furthermore, assuming that there's a single authoritative frame of reference whose very existence defines reality sounds a long-winded way of saying 'God'. OTOH, if we take the many-worlds interpretation, it pretty much reduces science to a self-selecting fallacy. That which we know as 'scientific fact' is simply what we've chosen to observe, and the lack of refutation simply means that the refutation happened in some other line of authority.
Granted, I don't know enough of the math to do the really heavy lifting that might give me a better perspective on the issue. But I've also never heard any discussion of what makes observation tick. AFAIK, it's just an axiom right now.
IIRC, Apple posted revenue of something like $7B last year, while Microsoft's was around $21B. I think the profits were also around a 1:3 ratio between the two. And I know that Apple made more money last year than Dell.
Given those numbers, and considering how new Vista is, I think it's probably wrong to say that Vista has brought Microsoft more money than OS X has brought Apple, at least so far.
Apple isn't in the business of "giving you what you want." They're in the business of "selling integrated solutions."
You don't want an integrated solution? Fine. You don't want an Apple product. It's as simple as that. Rolling your own solution is great, but don't complain that you can't cherry-pick pieces from Apple's integrated product stack to play with.
Name three other categories of consumer electronics which allow you to mix and match the way you can with a computer. Nobody who buys an HDTV complains that they like the display of model X, but would really prefer to load the user interface from model Y onto it.
For that matter, name three other product categories, period. Nobody complains that they can't use BMW parts in their Camry. Nobody questions Ben & Jerry's capacity to serve the market because "I want the fruit from Cherry Garcia, but I want to put it in Hagen-Dazs chocolate chocolate chip."
The vast majority of the market wants integrated solutions, and turning everything into a pick-and-mix festival does not serve that very large chunk of the market. Most consumers don't want to be drowned in a sea of options.
Yes, vendors can push inventory onto retailers.. it's called 'flooding the channel'. Microsoft may have done that with Zunes to meet its projected sales goals, based on its sales numbers per month for the year.
There are two reasons why Apple probably hasn't done so, though:
First, you can't flood the channel continuously. About all you can do is collapse the sales you would have gotten next quarter into this month's sales report.
Say I'm a vendor shipping a product whose market demand is a million units per month, and I want to puff up my sales figures. I ship 1M in January and 1M in February, then flood the channel with 4M units in March. That lets me claim 6M sales for my first quarter. It makes a nice press release, but doesn't mean that I'll ship another 6M units next quarter. In fact, if the actual demand is only 1M per month, I won't ship any at all. I can do the same thing again in the third quarter and claim 12M in sales by September, hopefully snagging some consumer interest for the upcoming Christmas season. But by the end of the year, I'll still only have shipped 12M units.
Bottom line, though: nobody can flood a channel with even a significant fraction of 100M units if the product doesn't already sell damned well.
The second problem is inventory management: Apple gives Dell a run for its money in keeping the supply chain thin. Apple itself owns less than 2 days worth of inventory at any given time, IIRC.
But in the example I gave above, I needed 6M units by the end of March. If I also want to stay below 2 days of unshipped inventory, I'd need a factory that can produce 1M units/month for two months, then jump to 4M units/month for March. Then I'd have to shut the whole thing down from April through June and start the whole thing up again in July.
It's incredibly difficult and expensive to jerk a factory's production capacity around like that. You can't build the product without machinery, which is expensive and can't just be rented for a month (who's going to have that much equipment stiiting around idle?). You need labor, which requires training (another sunk expense of both money and time) and doesn't like to be laid off every other quarter. And you need components, which means your suppliers would have to be willing to deal with exactly the same problems themselves.
In a word: unlikely.
Actually, the theory I've heard is that SCO wants to get its hands on the private (protected) documents belonging to IBM's lawyers.
Normally, communication between a lawyer and their client is privileged. The other side can't see it and can't subpoena it, no matter how much they want to or how much it might help their case. You can admit being guilty as sin to your lawyer, and that admission will never be allowed to enter a court of law.
Thing is, the privilege only exists if that information stays strictly between you and your lawyer. If anyone else gets their hands on information that was supposed to be protected by client-attorney privilege, the court will assume that you've waived the privilege and are willing to have the information appear in court.
If SCO could show that IBM or its lawyers passed privileged information to Groklaw, SCO could ask the court for permission to see all of IBM's privileged documents.
That would essentially take the case back to day zero. There would be a whole new round of motions to define the scope of privileged documents SCO was allowed to see, more discovery, more depositions, another chance for SCO to redefine its claims against IBM, and possibly (a very remote chance) give SCO some evidence that IBM actually did do something arguably illegal somewhere along the way.
At minimum, it would mean tons more delay. And SCO would also get an excuse to unleash all manner of punitive crap on PJ for investing the time, effort, and skill necessary to blow their media-driven FUD campaign out of the water.
It's important to remember that SCO's FUD was working until Groklaw came along. Journalists were still writing "there's lots of smoke, so there must be some kind of fire," articles even a couple of years after Groklaw got started. It was only when PJ piled up a big enough heap of documents, cross-referenced the claims, and then started collecting evidence to show that SCO might not be telling the whole truth, that the media machine stopped dancing to SCO's tune. It wasn't until journalists started actually looking at the evidence and drawing their own conclusions, instead of just parroting SCO's latest set of talking points or paraphrasing whatever article SCO had astroturfed out through its shills, that SCO really started to lose its case in the court of public opinion.
3. Transfer my efforts to a line of work society values more highly.
That's how economies grow.
Back in the 1950s, companies had whole rooms full of people punching adding machines. In the 1900s, it was all done in longhand. Today, a single person with a spreadsheet can do more math in 10 seconds than all the clerks in the world could do a hundred years ago. A six-man construction crew with backhoes and dump trucks can do more work in a week than a hundred ditch-diggers and dirt-carriers could do in a year.
The people who were displaced from those careers didn't just curl up and die, though. They found other jobs. And since we don't have to pay a few million people just to keep the books in order these days, companies can sell their goods and services at lower prices and still make a profit. Consumers get better quality goods for lower prices, meaning the dollar they earned doing something more valuable than digging a ditch goes farther than it ever did before.
In fact, we can engage in types of business that would never be considered under the old system. Can you imagine trying to run something like the iTunes store or Amazon with a bazillion human clerks crunching the numbers and keeping the inventory straight?
I think you'll find that the labels are the ones that have set the regional limits and pricing standards. Apple is bound by the contracts the labels were willing to negotiate, and the labels didn't want to negotiate liberal contracts when the iTunes stores were first being set up.
Having to run multiple, mutually exclusive stores is probably a dead loss for Apple all around. There's the massive duplication of effort in making each store run and managing the inventories, there's the effort of barring people in one region from using the store for another region, and there's the dissatisfaction from customers who can't get the music they want if it's only for sale in another region.
Apple runs the iTunes store as a value-added service for the iPod. The more music that's available, and the easier the stuff is to obtain, the more value it adds. How could it possibly hurt Apple to run a single store for everyone in the world, with all the music equally availble to everyone?
Given the track records of the players in question, I doubt that an investigation will find that Apple were the ones who went to the negotiating table saying, "hey, let's waste a lot of resources and piss off a lot of customers by making a patchwork of regional stores, offering different inventories at different prices in each one, and making people in one region wait six months longer to get access to their store than their neighbors 50 kilometers away!"
Actually, Apple is a vertically integrated systems vendor. The hardware's where they make the money, and the software is what makes the hardware worth buying.
Meanwhile, Microsoft is more like a component vendor to OEMs than a software house per se. Most people don't buy Windows, they get it bundled when they buy a computer. Most Office licenses (Microsoft's only other product that generates a profit) are N-hundred-seat deals, paid for by companies whose IT departments lay their own bundled software installation over the OEM's bundle before giving the machines to the workers.
Microsoft's revenue for 2006 was something like $21 billion, most of which probably came from less than a thousand major clients. Apple's revenue for 2006 was something like $7 billion, most of which was from people buying individual devices. Neither business model is inherently 'better' than the other, they're just different. Each is profitable in its own way, and each has its strengths and weaknesses.
The place where Apple is really hurting Microsoft is indirectly, in the midrange hardware market.
OEMs like Dell and HP have huge sales volumes because they offer a tier of low priced machines. That $300 PC isn't bringing Dell enough money to make running the production line worthwhile, but it does draw customers to the brand, and it allows Dell to buy enough parts to get good economies of scale. Then Dell uses those same (cheap-in-volume) parts for its machines in the $1500-5000 range, and that's where Dell actually makes its money.
But now Apple is competing hard in the $1500-5000 price range, and every sale Dell loses to the Mac screws with Dell's ability to keep the $300 machine lines running. But without the those cheap machines, Dell can't get the volume discounts, which makes it that much harder for Dell to keep things going. Basically, when you've built your business on leveraging economies of scale, losing those economies of scale is a bitch.
Thing is, Microsoft doesn't care about the machine's unit price. It just sells Windows as a component, so Microsoft makes $N per machine whether that machine sells for $300 or $3000. In other words, a 1-2% loss in the midrange market hurts Dell a whole lot more than it hurts Microsoft, but dumping the cheapest 25% of Dell's total market share would hurt Microsoft a whole lot more than it would hurt Dell.
Back when Microsoft was trying to kill Netscape, Microsoft execs used the term, "cutting off their oxygen." In the long run, midrange hardware sales are the oxygen for OEM's like Dell, and OEMs like Dell are Microsoft's oxygen.