The scheme does not applies to MS, which makes most of its Windows sales from bundling it to new computers after pressuring the constructors to ensure they won’t seriously propose an other operating system.
Lowering the value of a Windows upgrade would attract the value of the bundled version of Windows to lower levels, which would result in lower revenues for MS, because the volume won’t change (or the change will be too small)
Software is too expensive.
Upgrade paths in particular are utterly pointless (I know a few exceptions but I can count them on one hand).
So yes, SW pricing is all wrong.
One thing to factor in is the cost of support: when you have many customers, you also have much greater support needs. I think this is part of the reason companies like products that only sell a few copies for an outrageous amount of money: the support infrastructure required is minimal.
On the other hand, I think people have much lower expectations when they only pay $10 for an application, and you might be able to miff on support a bit.
So, I think your article is probably spot on–the most dangerous place to price your software is middle-of-the road. You get enough customers that you seriously need to consider your support infrastructure, and those customers typically have high expectations for the product.
From an economic theory standpoint, the example is confusing and misleading.
First, because it makes no practical sense w/r to market equilibria: software is essentially a zero marginal cost business, so sales revenue ~= profit, so pricing to revenue maximization should be easy. The first time anyone held a 50% off sale and their profits tripled, why would they ever go back to selling at the old price?
Second, because it’s an extremely odd way of presenting (and thus thinking about) the results. Following the “curve” thus implied, they should offer the software for free and make infinite profits. The usual way of thinking about this is as price vs. “quantity” (unit sales volume), and there is some maximizing price P* so that the area P*Q is maximized. When you make both the independent and the dependent variable in the graph be a function of the price, it’s confusing.
In short, this seems anecdotal and I’m skeptical that it is indicative of any bona fide industry overpricing of software. There are lots of other psychological/non-quantitative effects that go into pricing and could lead to an outsized sales effect (ex: consumer threshold behavior – there are certain magic price thresholds where sales suddenly take off, like dropping a piece of consumer electronics below $200) but most of those are very well understood by the industries that design products for them.
In short, there is obviously a large monopoly rent component to Microsoft or Apple charging $100+ for an OS upgrade. But monopolies extract rents because they are maximizing revenue, not because they are setting their prices too high. Are we seriously to believe that there aren’t whole buildings full of people at Microsoft devoted to fine-tuning those price numbers?
also have to poke your nose in at market saturation. Microsoft has reached just about as many computers as its going to reach (exept for perhaps a few people who might want a second computer next to their Mac or Linux boxes.
Microsoft has who its going to have, and most new computer users will end up using it anyway, no matter the cost… so why would they bother cutting the price? they probably wont sell many extra copies, and certainly not enough to justify.
On the app pricing topic there is definitely a downside to this race to the bottom. People are becoming accustomed to getting something for virtually nothing. Once established that expectation is almost impossible to reverse. i see a day when any application, no matter how complex, will be impossible to sell into the consumer market at a price exceeding $19.99.
Operating systems are no different. Because they come pre-installed and are not shown as a separate cost on the invoice, people think they’re free with the hardware. That makes it difficult to sell upgrades. Why should I pay for an upgrade when I didn’t have to pay anything for the OS in the first place?
Nothing Microsoft does can alter that perception. People expect to get a free OS with their hardware and no amount of upgrade discounting is going to change that. What’s even worse for Microsoft, and the rest of us, is that people are scared of upgrading their OS even when it’s free. A large fraction of PCs are running the operating system they originally shipped with, devoid of any updates. It’s no wonder malware is running rampant.
Of course you love paying for software, you are a software developer, that’s what in spanish we call a “professional deformation”, you are so caught up in your own profession that you see everything from a developer’s point of view.
However, from the consumer point of view, I don’t think it is ok to pay that much for something that ethereal.
I’ve always believe that there’s value in ubiquity, even to the point of giving stuff away for free.
It’s also worth noting the fact that when you sell more copies of whatever, you just plain get more users.
I’m not sure how valuable that is, but I suspect it is at least somewhat valuable. In the case of a multiplayer game like Left 4 Dead or Team Fortress 2, more users directly translates into a better overall game experience for everyone, so that’s definitely valuable.
Those Steam sales get me every time.
Definitely agree. In the specific case of Windows 7, I have 3 XP PCs that I’d like to upgrade, but I don’t have to. At current Windows 7 prices there’s no way I’m going to, so I’m going be waiting until it’s time to replace them. If there was a “why not” price MS would have my money in the bank right now.
Maybe software (with the App Store being the poster child) will soon be faced with the same problem that hardware and bandwidth currently have: prices approaching zero. Simply increasing volume with deep discounts will not work if the price is too low. A major paradigm shift in the business model for selling software will have to occur.
Chris Anderson covers all of this well in his book ‘Free’ (http://www.amazon.com/Free-Future-Radical-Chris-Anderson/dp/1401322905/ref=sr_1_1?ie=UTF8&s=books&qid=1249586583&sr=8-1). The psychology of Free and how you can make money with Free are intriguing topics.
Revenue is maximized when you set the price to the place on the demand curve where elasticity = 1. I’m sure Microsoft has done their research and knows what they’re doing when in comes to pricing.
+1 for the bargain effect.
That, and I want to see if Recaptcha expects me to type the umlaut over the “u” in one of the verification words I see now.
If this comment gets through, it didn’t expect me to. 
“I attribute most of today’s problems with the allowance of the Right-On-Red laws”
Wow, this is a new level of stupid.
I noticed the math error as well. The base price of $4000 was included in the 35% increase, but not in the others. The key word here is increase. An increase of 245% means 3.45 times the amount, not 2.45 times.
In the Valve case, for example, it’s hard to tell how many of the people who bought during the sale would have eventually bought a copy at the higher price. We can say for sure that anybody who bought during the sale won’t buy another copy later. So how much of the number is increased sales, and how much are just time-shifted, and they were simply eating from what would have been their long tail? It’s very hard to tell.
I think your view is distorted by the type of the software you are looking at. Why not sell it for free and make it up on profits is my take: http://devcomponents.com/blog/?p=350
The same Valve sales-trick is used with expensive clothing.
One of my friends: "I made such a good deal on these jeans! They were on sale for 80 euros instead of 120!"
Me: "Why don’t you just buy them from [some other store] at 25 euros?"
Him: "But these were on sale!1!!!111"
Me: "That’s calculated into the price, they expect to sell it for the sales price and it’s still really expensive, and they make a nice profit on it."
Him: “So what, I still made a good deal with the sale.”
He just likes to throw money away I guess 
Using the Valve model, I have actually been buying software I NEVER USE MORE THAN ONCE!
I have purchased about 30 or so games (some really old, some brand new) and play each one for a total of about an hour before I give up on it and go back to Day of Defeat: Souce.
Still… they have my money, and I think I ahve spent less than $100 all up for they lot.