The Dangerous Superficiality of the Software Marketplace
Imagine a professional product that a number of specialists use to do their work. It can be a niche product – let’s say it has just 1000 users, and that they typically open it twice a day. It takes 30 seconds to load all their data, they do some work, and then they’re done. (In other words, it’s much smaller and simpler than Tinderbox or Photoshop or Firefox – it’s a small niche tool.)
Now, suppose you’re the product manager. You suspect that you can improve the load time a bit if you have someone spend a day polishing it. Is it worth the money to save, say, 10 seconds of load time?
Let’s do the math: 1000 users, 10 seconds, twice a day. That’s 20,000 seconds/day. Over a year of 261 weekdays, I believe that’s 1450 hours. At just $25/hr, we’re saving the users about $36,000. At $100/hr – with salary, taxes, benefits, office space and furnishings not as impressive as it sounds – we’re saving $144K. A nice return for a day’s work.
If the product manager and the developer went off to Vegas instead and came home with $35K each, they’d be pretty happy and might take the rest of the department out for a steak and a beer. So they’d be happy. The customers still have half the savings in their pocket – and they can use the extra 1450 hours/year to make more stuff and earn even more money.
But the marketplace doesn’t care. New customers don’t even see the load time, because new users don’t have much data. So this kind of improvement doesn’t help you sell the program at all. (It might help if you were in a feature war, I suppose.) The technical press veers between incompetence and corruption, and certainly has not time for details like these. Given the choice between saving the users $36,000 or having someone spend the day polishing the layout of the website, the rational product manager probably says, “let’s go with the web site.”
The result seems to be a slew of inexpensive products with polish and limited capability, products that demo well and don’t attempt a lot. We have products like the impressive new Garage Band for the iPad that exist to demo other products. We have buzzword-compliant B2B products that promise vague and indirect benefits. And we have a rush to completely unproductive forms of software investment, like the billions of dollars being poured into buying invalid patents that are intended only to deter litigation from the owners of other invalid patents.
The disconnect between the real technical press – Gruber, TidBits, A List Apart, Tim Bray, Martin Fowler, Rands, Shawn Blanc – and today’s magazines and newspapers is nearly complete. As Gruber has long been showing us, there’s no penalty for being wildly wrong in Newsweek or the WSJ; you’ll be invited back to get another big payday to be wrong some more. And there seems to be no penalty for being a jackass – not merely being wrong, but being wrong in a way that confirms your prejudices or enhances your stock portfolio. There’s no penalty at all for being wildly wrong in writing reviews in the App Store. Meanwhile, the bargain basements of the Kindle store are overflowing the scanned “classics” so badly OCR'd that you can’t even be sure they’re in English.
We need to find a way to pay that fellow who is going to save us $144,000 this afternoon. We can use that money. To begin, let’s start penalizing the liars.