Let's stop wasting US$ 78 billion a year.

October 20, 2001 11:38 AM

Let's stop wasting US$ 78 billion a year. Is software development really this inefficient? Aside from the main theme, there is also an interesting statement from a CIO towards the end of the article. "Those folks [involved in the open-source movement] are very knowledgeable, very good at what they do, and they're producing really great code," [...]
posted by HeikoH (5 comments total)

That article was passed around here a bit. It's extremely accurate. I've been convinced for many years that support is the real reason why so many vendors hate open source. It's not because opensource projects have some sort of magic capacity for producing better code - that's just a function of good programmers and time. A business can easily get match an open source project because it can pay talented programmers to work full-time. Most businesses choose not to - they're happy to deliver marginal products with marginal support and don't want to spend the money improving.

As an example, consider Cold Fusion vs. PHP. I've built large sites with both and found PHP to far better on almost every count (speed (often at least an order of magnitude faster), features, flexibility, portability, better language, etc.). Unfortunately, one of the main problems to overcome when selling PHP was the "Open source means you're on your own" FUD most salespeople like to spew. This was very annoying as it was very strong opposite to my experiences.

I regularly found bugs in Cold Fusion on a regular basis (some features didn't work as documented and the server was generally unstable) which Allaire charges you to report. After getting no response back from their support people about a major problem with nested CFLOOPs, I posted a message to the user forums and was quickly told that this was a known bug that had be around since at least 4.0 (we were using 4.5).

In contrast, I found a total of one major bug in PHP that was similarly preventing a site launch. This was in the 4.0 pre-release for Win32 and the windows port was still somewhat experimental at the time (I had to use it because the client had pulled a "Did we say Linux? We meant Windows?" less than a week before launch). Somewhere around an hour after I posted a message to php-general, I had a patched version courtesy of one of the lead developers.

I've had similar experiences in most other cases (e.g. Apache and IIS may be close on speed but Apache wins by being portable, more reliable. significantly more feature-rich and much, much, much easier to manage). There are a few exceptions who provide good products and excellent support (I used to work at one of them) and I would like to stress that I'm not opposed to paying for software - to the contrary, I don't have a single program on my system which isn't paid for or free. What I'm opposed to is the attitude most vendors have - you have to pay for their product and when it breaks, you have to pay them again to fix it. What other industry gets away with treating customers like that?

It's bad enough to the point that most of the sysadmins I know tend to favor an open source program over its commercial competitors simply because while the open source products may lack the slick ad campaign they're almost always more reliable and easier to manage. That significantly lower cost of ownership really outbalances the need to do the occasional recompile or reading a man page.
posted by adamsc at 1:28 PM on October 20, 2001

The big problem is the disconnect between the domain in which software is bought and sold, and the domain in which software is written, used and maintained. I worked on a project a couple of years ago, in which the management were sweet-talked into buying a closed-source webmail application that we essentially had to reverse-engineer and kludge. When my manager met the people who'd written that app, they apologised, and said that he'd been misled about its capabilities by the sales rep.

Commercial software, at its worst, puts tons of these obstacles between programmers and users: quite simply, the aim of many big software companies' sales departments is to perpetuate the upgrade path, and make customers pay for features they won't need or support contracts they won't use. I do suspect, though, that the "newest! latest! fastest! least tested!" culture is starting to run out of steam, at least when it comes to large-scale buyers.
posted by holgate at 2:45 PM on October 20, 2001

?

/?
posted by greyscale at 3:39 PM on October 20, 2001

"Those folks involved in the open-source movement are very knowledgeable at what they do, and they're producing really great code."
—RAYMOND DURY, CO-CIO OF AMERITRADE

Hey, thanks Raymond! Flattery will get you everywhere. I'm switching online brokers!

It's interesting that so much is being made of this story, although it is hardly something new. Take a look at what people in the software industry have been saying for years. Guy Kawasaki says "Don't worry, be crappy!" Richard Gabriel explained why "Worse Is Better"... and frankly, from a software business perspective, both are right.

In my prior experience doing QA in the software industry, we had the thumbscrews put on us numerous times to ship software that we didn't think was ready to go out the door. Talk to anyone in QA... the pressure is definitely there. All the 80 hour weeks leading up to a major release don't matter, because if there are still bugs in the software, it is obviously your fault for them being there, right?!

The answer, I think, isn't just open source. The answer is for some sort of open source group to work with companies to not only adopt open source software as a standard within x amount of time, but also to dedicate money to hiring developers to improve the software they use. If these two elements could be combined somehow, a lot more progress would be made and everyone would benefit.
posted by insomnia_lj at 3:05 AM on October 21, 2001

I think the difference now is that this story covers the issues at the CIO level. In the past, the CIOs of the world were generally not aware/concerned of the problems in the software they'd purchased (after all, the sales guy assured him it was great over that four-figure dinner) and might even have blamed their staff for the implementation problems. This lead to people sneaking in other software or, if possible, not using the software at all - all quite unofficial and while it made their lives easier, it doesn't hurt the vendor.

If enough top-level managers realize how their traditional trust of the software and hardware vendors has been abused and start demanding better quality, we might see changes happen much faster than if the changes creap in from the bottom. A single large company announcing that, say, it will be using StarOffice 6 instead of Office XP would get a great deal of from Redmond.
posted by adamsc at 10:46 AM on October 21, 2001