Chad Perrin: SOB

12 December 2006

the software price:quality curve

Filed under: Cognition,Geek — apotheon @ 01:09

Michael Covington makes an interesting point about the relationship between price and quality of software. Specifically, he talks about how in many tangible-product markets it’s usually a good idea to buy stuff from the mid-price range, where you’ll get the most return on your dollars in terms of quality, unless you have very specific needs only satisfied by the most expensive offerings. The unspoken corollary to that is the fact that you might also need to buy one of the cheapest, if that’s all you can afford and you simply need something without much concern for quality. He then goes on to contrast the software industry with this price:quality ratio curve.

He points out that, anecdotally, he has satisfied himself that the best software is in the outlying ranges. That is to say that the free or cheapest software and the most expensive software tends to be the best, while the mid-range is full of crap. His theory is that the cheap/free stuff is a labor of love, and thus truly inspired and high quality as a point of pride and enjoyment of the craft, while the most expensive stuff gets there by being good enough to support the prices.

What Michael Covington says is sorta true . . . except that software exists in a monopoly market. This throws a monkeywrench into things.

I’m not actually talking about Microsoft this time, believe it or not. I’m talking about the fact that proprietary closed source software is the usual form taken by mid-priced and expensive software. This doesn’t change the fact that mid-priced software will usually be lower quality than cheap/free and expensive software on average. I do agree with that estimation of conditions. The problem that arises is the simple fact that when you start getting into expensive software the curve falls apart in a monopoly market.

“Intellectual property” is, by definition, a monopoly for a (nominally) limited time. That monopoly allows vendors of pricey software to leverage real market monopolies not just in the software itself, but in software within the same class. This leads to high-end software being of very unpredictable quality: it is neither as predictably good as free or cheap labors of love, nor as predictably bad as mid-range purely money-making software without inspiration.

A man named Kevin Korb, in a LUG mailing list discussion, brought up the multimillion dollar range where often “enterprise” software is terrible quality, so atrocious it’s difficult to imagine anyone wanting to spend even two dollars on it, let alone two million. This is amply covered by my above modification to the theory of software quality correspondence to price: only software that achieves an effective monopoly on a given market niche is capable of rising to that price range, and with software that has such a monopoly there is little or no motivation to maintain real quality in the software. Instead, the motivation is entirely to keep money rolling in with support contracts and upgrades. Both of these revenue streams are dependent upon dissatisfaction with the current state of things, to a certain extent, which actually provides a very strong disincentive for producing software that is “too good”.

In other words, from a purely economic forces standpoint within a monopoly market (the commercial software market is currently almost entirely predicated upon the intellectual property laws enforcing nominally limited monopolies), the response to the notion that the absolute most-expensive software is some of the crappiest is “Of course it is!”

There’s an additional factor at work in determining the quality of free software as well, of course. When it’s that cheap, and that free of encumbering monopoly enforcement, it must stand or fall on its own merits. The underlying assumption of the U-curve analysis is that the cheap/free software is established and recognized to some degree: this is dependent upon people having used the free software, liked it, recommended it to others, and thus contributed to its growth into an established piece of software. In other words, it’s not just the fact that it’s a labor of love that makes the free stuff good. It’s also the fact that, by the time you’ve heard of it, it is proven in the test of evolution, on the battlefield of the survival of the fittest.


  1. i think in addition to price:quality you also have to consider price:value. the reason companies will shell out $millions for crappy ERP software (requiring more $millions for maintenance) is that it’s essential in order for them to make $100 millions and $billions. these products do have a monopoly on the market, but it’s not because no one else could write the same kind of software – it’s because nobody wants to unless compelled by a significant salary.

    also, what about free software being provided with a pricey service contract, a la red hat?

    when it comes to software, you can’t simply look at the software itself. there are all kinds of confounding factors around it – support, integration, compatibility, training, maintenance, principles, etc. – that determine the actual value of the software to someone; and then, there’s the perceived value, which is used for purchasing decisions but frequently has no relationship to the actual value.

    Comment by sosiouxme — 13 December 2006 @ 07:10

  2. Many times corporations will go with the high-priced package for a sense of security. It feels better to them to pay a lot for an “enterprise” solution, even though it probably isn’t as good as a freeware version of the same thing, because they can’t imagine that anyone could ask that much money for crap. And if they did, they couldn’t succeed (goes the internal dialogue). But they do, and they do.

    Comment by Sterling Camden — 13 December 2006 @ 05:56

  3. sterling, there isn’t a real alternative. i know there are open source “enterprise” apps in various stages of development – i wish them well, and i’m sure they’re adequate for many purposes. they’re not for a large enterprise, and i am not optimistic about open source reliably keeping up with the many tedious business details that such commercial apps do.

    Comment by sosiouxme — 13 December 2006 @ 10:20

  4. You say there is no alternative for the “large enterprise” applications that cost in the millions. Please provide examples.

    The only cases that come immediately to mind for that are not, in fact, enterprise application packages at all. Instead, they are custom software solutions, contracted and supported at a cost of millions, particular to the specific enterprise in question. I’ve never heard of any packaged software in that price range more than a couple years old that didn’t have credible competition.

    I’m not saying it doesn’t exist — only that I haven’t seen it.

    Comment by apotheon — 14 December 2006 @ 02:11

  5. perhaps we’re thinking of different things, then. i was thinking specifically of the Oracle Financials or SAP applications. at that scale, where you buy only one instance per enterprise, it’s kind of a given that it will be “customized” by a raft of consultants. can you think of anything in the $millions range that doesn’t require an army to integrate, customize, and maintain?

    because there’s one to a customer, the system is so integral to the customer, and each customer has so many subtle differences, nothing is going to be a complete solution out of the box.

    Comment by sosiouxme — 17 December 2006 @ 09:13

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.

All original content Copyright Chad Perrin: Distributed under the terms of the Open Works License