A Journey Through RISKS

A day in the realm of Jamie Zawinski and the world can never be the same.

Get It Try It

It all started when someone sent me a link to Jamie Zawinski. It had something to do with the continuing demise of Netscape, something which I as many others really mourned. Jamie was in the show from the beginning, and now he was getting out. He'd seen the beginnings of Moz, he'd followed it along a while, but gradually he began to realize that the magic was gone. He got out.

Reading through Jamie's chronicles is a saddening experience. Jamie really burned out over the whole affair. He spent a great deal of his career and his life at Mosaic, later Netscape, and he was evidently hurting bad. Jamie has a lot of links at his site; I started clicking around and somehow ran into Eric Raymond.

Eric is the author of The Cathedral and the Bazaar, an analysis of a project called fetchmail. The book purports to prove that the open source approach to software development wins hands down. Even here I found links to click; this time I ended up at a web forum called The RISKS Digest.

The RISKS Digest is an off-shoot of the venerable ACM. It's a moderated forum, run by Peter G. Neumann. Every two weeks or so P.G. assembles the mail he's received and puts it together into a new edition of RISKS. The RISKS Digest reaches perhaps more computers than any news group in the world; it is certainly read by people who matter. Reading through the articles there, I realized I was in very distinguished company indeed. Then something came over me - it's hard to describe what - but I waxed very nostalgic, for the days of yore and of UNIX and its elegance, and I started searching the site for an email address, for a link that would tell me how I could submit an article of my own. It wasn't easy to find, not for me, but after considerable effort I found it, went offline, and began to put my thoughts down on electronic paper.

You can read the article here. I had no idea this article would have the audience and the effect it had; I had no way of even expecting that something I penned so hastily would make its way into this select forum. But it did, and it evidently caused quite a stir, and you can find it today at sites all around the world, from MIT to Australia. It also resulted in quite a lot of email.

Most of the email was highly positive and complimentary, but not all of it was. One of the first letters that arrived came from Microsoft. In my article I mention RegClean, an extreme eyesore on the bloatware map, and now I was getting lambasted by an angry person with a suspicious email address out of the Microsoft domain. Frankly, I was scared and intimidated.

The Microsoft letter made some very sweeping and extremely cheap claims, and the best way to counter such claims was to take the object in question, RegClean, dissect it completely, and publish the results.

The result of this 'scientific' study you can read here. This article caused, if possible, an even bigger storm than the first article. This one got onto all sites world wide that the first article had and then some. Very nice people from all over wrote to me and asked my permission to publish the article on their own forum sites. I had somehow managed to strike a very welcome chord.

The London Daily Telegraph contacted me several weeks later. They were planning on devoting their coming Thursday 'Connected' supplement to the subject of bloatware; someone there had seen my dissection of RegClean and wanted my opinion about new Microsoft statements on the subject.

Somewhere towards the beginning of the exchange I mentioned that we were working on a new program, which for want of a better description I said was a kind of 'Explorer' but without the bloat. I also mentioned that the project was going rather well and that the program currently weighed in at twenty six kilobytes on disk. After an hour or so we hung up and I waited for Thursday to see what the Telegraph would write.

You can read the Telegraph article here. What really took me by surprise (there were a number of things that 'took' me) was that the author of the article, Mark Ward, was not just a reporter, but an authentic bloat crusader who'd evidently paid some dues of his own. What surprised me further was that Mark had been so honest throughout. Mark didn't write the article to sell copy; he wrote it as it was, exactly as I'd said it.

I was also surprised to find out that Microsoft ran undercover propaganda sites like Slate, which exist solely to spread propaganda about the Redmond company. And finally I was surprised to read that radsoft.net now had an Explorer killer in the works that weighed in at 'a svelte 26KB'.

We who have written software for radsoft.net have never really gone out of our way to shrink applications; we've never resorted to compression techniques; lean and mean has not been an end-all in and of itself; we've just happened to produce, by way of comparison, extremely compact industrial strength programs. As we've said so many times, it's not we who do the unusual stuff, it's the other guys, the guys who churn out megabyte executables without even blinking. Their thinking is what's extreme, not ours. There are so many people coming into the world of the PC today that don't realize that: they've never been given the chance to compare. By the time they arrived the market had been ruined.

But now it was out there: I'd made claims, and now I was going to have to prove it could be done. It was no more and no less than 'put your money where your mouth is'. Mark Ward had quoted me on our X-file project. Neither I nor anyone else had the slightest clue whether this application would make it into port at 26KB. It might end up 24KB or even smaller; it might end up bigger. But now it would have to make 26KB come rain or come shine.

The half year that followed was the most gruesome in my career, in my life. X-file, as the program is called, had been my 'baby' right from the beginning. I was the one who first sounded the alarm. I was the one who had seen how Win2K was trying to undermine use of NT's File Manager. I was the one to discover that Win2K wasn't even shipping File Manager any more. None of us had ever used the Explorer - we hated it outright - and we weren't going to start using it now. X-file was sketched out some years earlier and then just abandoned, for lack of ambition and for lack of need.

But now it had to be done. As early as April I'd really and truly thought I was finished. I hadn't written a piece of software from scratch in several years: we'd all been out teaching, and loving it too. Great to get away from that frikkin monitor, so much better to do the classroom aerobics and have a real ball traveling around the world. This programming stuff was gruesome! But it had to be done. And the results had to be published in RISKS, and the program had to be made available to all who asked, as this was the proof submitted: that bloat is an engineering crime, that bloat need never be condoned, that bloat is a mindset that can be avoided. Had I failed by one kilobyte on the image size the vultures would have been all over me. After a brief and very enervating period where the image size grew up to 28KB there came a respite where suddenly all the pieces of the puzzle started fitting into place and the image size came back down again, almost by itself.

I went public with X-file on 16 October 1999. You can read about it here. It didn't really cause a storm, it resulted in a lot of requests for the program, but it did take a lot of pressure off me, and in my mind proved my point and the point of the bloat crusaders once and for all.

X-file is a complete file system browser: it matches nearly every function in NT's File Manager and Windows 4's Explorer, and even adds a few cute twists of its own. Yet these latter two weigh in at a quarter of a megabyte on NT, and X-file is merely a tenth of that size.

There are other reasons besides bloat for the dramatic difference in file sizes: both NT's File Manager and its Explorer are Unicode applications; X-file is not. Both applications can act as OS shells; X-file cannot (yet). But even these considerations do not account for much of the difference. Almost all of the Explorer's shell technology is in shell32.DLL; File Manager's shell capabilities do not cover the features showcased in Windows 4.

The one thing I have learned from it all is that the little I'd known, the tip of the Titanic that I'd seen, is all very commonplace to the regular readers of RISKS. They've watched Microsoft and the commercial software industry from afar all these years while I had somehow lost that perspective. I'm thankful to the readers of RISKS for being so helpful and for showing me what life outside is like - PG and his friends are a rare breed indeed.