Bastion Products

Information

What is Bastion?

Bastion is an upstart start-up software developer dedicated to the notion that good software can be inexpensive and useful software can be small and efficient. We are primarily focused on development for Mac OS (mostly OS X at this point) and PalmOS with occasional interest in BeOS and other POSIX-ey systems. You might be able to talk us into Windows with a lot of money, but generally Windows development has a documentably lower ROI than other platforms and it's not worth the hassle.

Who is Bastion?

Bastion consists of two programmers, each of whom has several years of experience with multiple platforms and languages. Collectively, we're comfortable with BeOS, Java, Mac OS (traditional and OS X), several Unices, a few flavors of Windows, VMS, Apple DOS/ProDOS, PalmOS, and some others you've probably never heard of. Most of our development work is in C++, and we are firmly dedicated to the notions of portable code, strong UI compliance, and efficiency.

What's the beef with Windows?

Like I said above, it's not worth it to me. It's harder to make money in the Windows market. Here's why:

1. Capital Expenditures
   

Proper equipment for Windows development costs more to buy and support over a given span of time than comparable equipment for other systems. This is especially true in the case of Mac OS, but even holds to some extent for other traditional x86 systems due to Windows' relative volatility. Note to platform bigots: No it's not true that Macs inherently cost more than x86 machines. Historically the prices of truly comparable machines have been quite close with, if anything, a slight edge in favor of Macs. The apparent discrepancy is that you can get an x86 machine for less money than the contemporary low-end Mac. If you think that matters, you're confusing cost with value.

1. Development Costs
   

Worthwhile Windows toolchains cost substantially more than gcc or Apple's MPW (historically) and XCode. If you want the best possible Windows compiler, you have to pay Intel for a plugin that only works in Microsoft's development environment. MPW, gcc and XCode are free. MetroWerks CodeWarrior for Mac OS isn't free but comes with one of the best C++ application frameworks (unfortunately platform-specific) I've ever seen. Windows programming documentation is abysmal. Take a look at the store shelves. See the racks of competing Windows programming documentation. Part of the demand for such things is due to the difficulty of finding complete, coherent documentation at any price.

1. Per-unit Support Costs
   

As with the capital expenditures argument, this is mostly — but not exclusively — in comparison to Mac OS development. The fragmentation of the Windows hardware and software base and the relative lack of real or de facto UI standards raises the cost of after-sales support significantly. Windows is not Windows; there are several Win32 variants in widespread use, always a new one growing. It's entirely possible to write technically correct software that only works for a small portion of the user base.

1. Viable Lifespan
   

I've commented twice in the lack of uniformity in the installed base. This also translates to an increasing rate of failure of correctly coded software.

1. 3rd-Party Competition
   

Vendors who don't do a lot of analysis will go Windows by default. I'm not saying here that vendors who do perform substantial analysis won't do Windows, just that it's also the knee-jerk reaction. As a result, in most product categories, the user has far more choices than they would on other platforms. For any individual vendor, this is likely to be detrimental. Depends on how big the pie is and how big your slice is.

1. Shelf-life
   

Because of the large amount of competition mentioned above, some market segments have an unbelievably high turnover rate at retail. Six months is ancient history for all but an exceptional few in the game market, for example.

1. 1st-Party Competition
   

Windows users are the only market a 3rd-party vendor could target where you're likely to face substantial competition from the platform vendor. Let's face it. You could produce a word processor that flies on a 5-year-old machine, fits on a floppy, never crashes, does everything 90% of the word-processor market needs and charge $100 for it and you wouldn't recoup your costs because you'd be competing with Word.

1. Per-capita Purchasing Habits
   

Another argument that's more pro-Mac OS. It has been documented repeatedly by such bodies as the SPA that Mac OS owners buy about twice as much software after-sale than Windows users.

But what about Windows' dominant market share? Yes. Microsoft's share of the retail personal computer operating system market is massive. But market share really isn't a legitimately interesting number to anyone who doesn't have a financial interest in the vendor. Okay, fine. Windows is preinstalled on 9 out of every 10 machines sold today. Does that mean 90% of the world is using Windows? No. Try to buy a pre-built x86 without Windows. Go ahead; try. It can be done - and it’s much easier than it used to be - but it still takes more effort than most people would think it's worth. So how many of those boxes go home and get something else put on them immediately? Few, but some. A larger percentage of the custom-built machines in use don't get Windows put on them, and they're not counted in the statistics at all. Really, the infusion of new copies of Windows into the installed base is probably significantly lower than what most people are led to believe. Continuing, please note the use of the phrase "installed base." See, the potential market for 3rd-party software vendors isn't the market for operating systems. It's the installed base at the time their software ships. Market share represents the portion of goods sold in a given market in a specific period of time (usually a quarter) that are attributable to a given vendor. What isn't represented in that figure is what the installed base looked like at the start of the period and what the relative rates of attrition were over that period. Here's an example:

A and B are the only two foo vendors in a given geographical area and their foos aren't plug-compatible. You are one of several vendors of foo accessories and need to decide which company to support. In the most recent quarter, A sold 45 foos and B sold 5. A's market share is thus 90% and they are obviously the foo vendor with which you want to be compatible, right? Hold on a second. See what you forgot to notice was that when the period started there were 500 foos from each vendor already in the field. You also didn't notice that while A was selling all those foos, 42 of them were replacements for existing A foos which had broken down. One B foo was a replacement. So what's your market? Not 45 to 5, but 503 to 504.

Let me sum it up with a real example: In the 5 years before this was written, 5 copies of Windows have been bought on my behalf and not one has contributed to the installed base. In that same time frame, I've increased the Mac OS installed base 7 times.

So what does tick you off?

I think it's best to illustrate by examples. Policy statements tend to be either to narrow or too broad to be effective. This list may grow:

1. A few years ago, the local newspaper published a machine review by a syndicated author. I don't have the machine in question, but I am familiar with it. The stock configuration has 3 free PCI slots, 2 open IDE channels, and ports for USB, FireWire and legacy keyboard and mouse. 100Mb Ethernet and adequate sound I/O were on the board. One of the first things the reviewer said was that the machine wasn't expandable. Not "not expandable enough for my needs" but simply not expandable. That's sloppy at best. Worse is that I wrote to the paper and the author to point out the error. No retraction or correction was ever published. Note to experts: If you're reviewing something, it behooves you to actually use it...or at least look at it. You do a disservice to those who don't know better and make yourself look like a fool to those who do.
   

2. That reminded me of a similar but distinct annoyance. I can come up with lots of specific examples, but basically I find it really obnoxious when people make a factually inaccurate statement and their response to a well-documented, polite correction is not to say, "Oh, sorry; my mistake," but to get belligerent and sometimes downright hostile in their support of an unsupportable argument. Note to experts: There's a lot of common knowledge that isn't actually true.
   

3. I once saw someone assert that a specific coding technique was "literally 100 times faster" than an alternative. Since this wasn't my experience (by a long shot) I asked him if he could document the claim. His response? "I just said it that way to emphasize the distinct improvement." (The improvement, btw, was a false one exhibited by a quirk of the way his compiler optimized things in a degenerate test case.) Note to experts: Don't say "literally" unless you know it to be true... especially as a way of being intentionally hyperbolic. There will be people who take you at face value and end up repeating it.
   

4. "Well that's my opinion, and opinions can't be wrong." This is, perhaps, one of the most annoying pseudo-philosophical debate tactics to come into vogue in public discussion over the last decade. It's a statement uttered by someone whose uninformed beliefs don't jibe at all with measurable reality. A clue for the clueless: An opinion is an expression of belief about a subjective matter. A positive assertion regarding an objectively verifiable aspect of reality is not an opinion. "That's a pretty flower" is an opinion. "That's a lily" is not. Labeling an assertion as an opinion does not magically exempt it from tests of validity. "That's a lily" may be a statement informed by a knowledge of botany. It may be a guess. But if you're looking at a rhododendron, you're not looking at a lily and saying "that's my opinion" doesn't change that fact or render it ambiguous.