Dr. Jerry Pournelle

Email Me

Why not subscribe now?

Chaos Manor Subscribe Now

Useful Link(s)...


Hosting by

Powered by Apache

Computing At Chaos Manor:
The Mailbag

Mailbag for January 8, 2007
Jerry Pournelle jerryp@jerrypournelle.com
Copyright 2007 Jerry E. Pournelle, Ph.D.

January 8, 2007

On the topic of the Microsoft Ferrari Giveaway:

I have a lot of years doing PR for computer companies (starting in early 1974 for Ohio Scientific Instruments, which lets me claim the longest-in tooth anti-award) but in recent years, reshaped the Newstips operation into being journalists to journalists; in either role, I deal with tons of companies and policies, and tons of media and ethical standards. I established a policy of helping anybody in the major media get any product from any company any time, so there are wives of guys in network news who refer to me as their husbands' "pusher". I generally find that the main reason for ethical outrage (short of the scandals we all read about) is jealousy; those on the receiving end seldom complain, as long as the presentation comes with no strings attached and no sleazy undertones, but those who wish to receive and don't get that visit from some PR Santa will often try to use ethics as a high ground to cover what is essentially a blend of disappointment and greed. Overall, few journalists ask for or accept anything that isn't a candidate for coverage, and as you note, it's generally a greater kindness for a reviewer to not share his or her disappointment in a bad product if it isn't amazingly bad. I've been doing a review a week for some years now, and I'm delighted to slam the genuinely disappointing stuff (like the phone sync software that corrupts your Outlook data files), but the yawners are not generally worth it.

My advice to PR people: learn to calculate the return on investment in actual revenue dollars. Empirically, we know that 1.5% of a self-eligible audience become instant adopters on hearing about a cool new product; in a Pournelle column, the self-eligibility factor is high, so despite a circulation footprint that's a lot smaller than, say, the Washington Post, the actual unit sales result for a highly technical product may in fact be bigger. What is the dollar amount per sale that's allocable to funding a PR effort? If, for example, it's a $300 product with a $10 allocable revenue allowance and coverage in a Pournelle column can bring 300 sales, a PR operation just created a 10:1 ROI by giving the product away - or, more appropriately, investing in its coverage. But good luck ever finding 10 PR people who understand this.

My advice to journalists: when requesting a product, be as specific as you can about the number of people you reach (not clicks, not page-views, not copies-printed, not market size) and in fewer than 10 words, what they get out of reading you. If you want a product on a one-way trip, ask for that; if they expect it back, they should expect to provide a return shipping account number.

Martin Winston

As I said last week, few reviewers can afford to buy enough equipment for review. As to returning equipment, in my case at least I tend to review by using something. If I use it and don't like it, I'll get rid of it fast. If I do like it, I'm not going to want to tear it out of the system and send it back until something better comes along if I've already gone to a lot of trouble to put it in as a main machine.

Most of us get far more stuff than we know what to do with. Literally. Giving it away generally doesn't work because the recipients usually want support which we can't provide. And most of it's pretty obsolete before we finish with it.

Last week I wrote about Vista and the future of operating systems. Peter Glaskowsky commented just too late for publication:

From the column:

"I'll go further. It would not at all surprise me to find that Vista is Microsoft's last operating system; after Vista they might find it prudent to design a good Presentation Manager for UNIX and LINUX and get out of the headaches of the OS business. After all, at one time Microsoft was designing the Presentation Manager for OS/2. It's hard to imagine Microsoft without the OS income stream, but it's entirely feasible. Virtualization and presentation are more important to the user than the nuts and bolts of the OS anyway."

You may remember that a few months ago I suggested Microsoft develop an all-new OS without low-level legacy compatibility.

That's essentially what's being suggested here, but I think that's not well-enough understood, so the consequences of the move are not appreciated.

Microsoft just released a new OS with dozens of new APIs. Most are small, but some are quite substantial: WinFX, which succeeds Win32; new communications and workflow APIs; and Avalon, which is the new presentation manager.

Unix is built on its own foundation. Microsoft's APIs were not designed to run on top of the Unix foundation, so it would be very difficult to make existing Windows apps run on top of Unix. Even if compatibility was achieved, performance would be poor, and stability and upgradability would be severely compromised. We've seen these problems before, every time someone tries to bolt together fundamentally incompatible software objects.

Avalon is probably the easiest part of it, which I suppose is why the proposal boils down to "Avalon on Unix"-- but Avalon is also the least important part of Windows. Windows developers have to use all the APIs, but their greatest commitment is to the APIs for file manipulation, network access, and so on.

If Microsoft gives up low-level compatibility, it loses its lock-in with the software developers. All of their Windows applications become dead ends. They could develop for the Unix foundation underneath the old Windows APIs, but APIs are valuable abstractions for software developers. So would Microsoft also deploy Unix- compatible APIs within this future OS?

Microsoft would have to choose between losing performance and stability, or losing control. Both choices would mean the end of Microsoft's dominance of the computer software industry. I know a lot of people would like that, but you can't seriously expect Microsoft itself to pursue such a goal.

My suggestion was that Microsoft should develop an all-new OS in parallel with Windows, allowing a transition from one platform that Microsoft controls to another that it controls. Personally I'd rather see some other company come up with the Next Big Thing, but the question at the time was about what advice I'd give Gates & Ballmer.

Microsoft has chosen evolution over revolution twice before, recall-- Microsoft could have adopted Unix or some other existing OS when it developed its first GUI, but it didn't. It probably would have been easier to adopt Unix instead of developing Windows NT, but that too would have been a mistake for the company.

In both cases, Microsoft took several years to discontinue the old platform. They didn't try to force everyone to switch from DOS to Windows 1.0, or from Windows 3.1 to Windows NT 3.1.

Apple, on the other hand, did try to force its customers to switch from the Apple II to the Mac, and it nearly killed the company. Now, true, Apple pushed pretty hard to move its customers from Mac OS 9 to OS X, which was based on FreeBSD, but that situation can't be repeated by Microsoft; it doesn't control the hardware makers, and its customers are not exactly emotionally committed to Microsoft per se (one way to benefit from having a small market share and setting yourself up as the alternative to the market leader).

. png

Robert Bruce Thompson commented:

As I see it, Microsoft has three choices:

1. Attempt to do another update of the original Windows NT code, which is what they did for Windows 2000, XP, and Vista. Given the amount of resources they devoted to Vista, which ended up a very pale shadow of what it was supposed to be, Microsoft simply can't afford to do this, literally. Vista cost them five years and supposedly upward of $10 billion. There's no reason to think that a follow-on OS would take any less time or be any cheaper, and many reasons to think the converse would be true.

2. Write an entirely new OS from a clean sheet of paper. This renders all existing Windows software incompatible, which obviously sacrifices their installed-base lockin and their developer relations. This is what they should have done instead of doing Windows 2000.

3. Write a GUI shell to run on top of Linux or BSD (given the GPL versus the BSD license, I'm sure they'd choose BSD.) Dedicate some programmers to update and improve BSD on an ongoing basis, and contribute those updates back to the community. That gains Microsoft good will from the community, and simultaneously ensures that they have a solid foundation for their GUI shell on an ongoing basis. I don't know enough about Windows internals to say how difficult it would be to port Windows APIs to a BSD foundation, but I suspect it's within Microsoft's power to do so. After all, the free WINE project has done a pretty decent job of it even though they're very limited in personnel and other resources and have no access to internal Microsoft documentation.

If I were Microsoft, I'd choose #3 and make strenuous efforts to get the Windows APIs fully ported so that most, perhaps nearly all, existing Windows applications could run on the new OS. I'm sure that there'd be lots of problems getting that done, but I can't see that Microsoft has any alternative at all.

Robert Bruce Thompson

Given the increasing use of virtualization - some magazines are calling 2007 The Year of Virtualization - it may not make a lot of difference what operating system an application runs on. With computing plenty and virtual machines, everything will run anything. The Intel based Macs with Parallels are only the beginning.

We have a great deal of mail concerning last week's letter on changing the location of My Documents and other such files; far too much to publish it all. Here's some representative samples.

Subject: Changing My Documents Location

Not sure why your MS correspondent this week didn't mention it (perhaps because it is officially unsupported) but Tweak UI directly supports changing the location of the various "Special Folders". The option is in the My Computer section. You can change My Documents, My Pictures, etc. I have used this on XP under Parallels to make sure all my "user" files in XP save onto a Parallels shared folder that is actually a Mac folder.

TweaK UI is one of the "PowerToys for Windows XP" found at this link.

Sean Keeley Toronto, Canada

And this:

Subject: Moving My Documents

There is a lot easier way to move My Documents to wherever one desires. Simply right click My Documents, select properties and there's a move button. Its in both XP and 2000.

I discovered it purely by accident.

Warm regards, Gary

And this:

Subject: moving My Documents etc

Re: Moving My Documents to another location:

Jerry: Your MS correspondent needs to look at some of MS's unsupported tools. Powertools for XP includes "TweakUI"; instaling the software, select the ""My Computer" menu, look at "Special Folders", and look at the dropdown list. You can reset any number of directories to your choice of locations, and not have to fiddle around with regedit at all.

As always, MS makes an esay way to do things, then refuses to tell anyone about it.. even their own people.


And this:

Subject: Windows My Documents Folder


Noel Nyman's response to the relocation of the My Documents folder, while informative, demonstrates the overall lack of a real architecture within Windows. The registry is a poorly documented mess and lacks the tools to allow a mere mortal to customize their system in a way that they might like.

While your Winword directory is fine for MS Office applications, many apps do not offer a way to change the default directory from My Documents.

Bob Holmes

On Virtual PC and the PowerPC Mac ...

Mrs. Roberta Pournelle is a certified Reading Specialist, and has taught over 6,000 people to read. She has taught in private schools, public schools, and as the teacher of last resort in the Los Angeles County juvenile justice system. Over the years she developed a program which can be used to teach both children and adults to read. See this link.

The original program was written for the Atari, then in DOS for the PC. That version required a tutor to read what was on the screen. Then Apple developed speech synthesis more than adequate for the task, and we used SuperCard to write a version for the Mac. Apple has since changed operating systems and that version no longer runs on modern Mac systems. Finally, we did a Windows version. The program was written in Delphi by Robert Ransom, and Roberta recorded many thousands of phrases, sentences, and syllables. This version will run in Windows on an Intel Based Mac, either in Parallels or using BootCamp.

We recently tested the program on a PowerMac with Windows XP running under Microsoft's Virtual PC. We had some difficulty in doing the required upgrades to both Virtual PC and then Windows XP, but we did manage to do that (it mostly takes patience). Once the upgrades were installed, her program ran quite well on Ariadne, my 15" PowerBook.

That experience generated this letter:

Dr. Pournelle,

It is just barely possible to run XP on a recent Power Mac. I used to do it when there was no other alternative. It had plenty of drawbacks besides the languorous speed. There were some Windows programs that it flat wouldn't run. I heavily rely on Visual Studio to do much of my work, and that was a no-go.

I purchased a dual-core Mac mini which I shuttle to work, and have been very pleased. I have partitioned the drive to use Boot Camp, but I never do because I run Parallels. It is, as I recall you describe once, flipping bliss. Visual Studio works fine, and I have had no problems with any Windows program I have used (which is an admittedly small set, but includes Office programs, OpenOffice, and IE6 and IE7). I have Norton System Tools, McAfee Virus Scan, and Windows Defender installed and they all run fine. I have read some analysis that there is a couple percent overhead, and it may not be up to FPS game, but for me it acts just like a PC box. Just for fun, I have had OSX, XP, and Ubuntu Linux running simultaneously.

As an aside, if you want to install Linux on an old PPC Mac (so that it can run a "modern" OS), Yellow Dog is the way to go. I am writing this on a slot-loading iMac running YDL, and it is perfectly fine for light email and web-surfing, including online banking, buying tickets, etc. I consider the lack of Flash plug-ins a bonus. Ubuntu is nice, but very slow. I've also tried SUSE and Fedora with unsatisfactory results.

I have some interest in your wife's reading program. I have kids and lots of Macs. I understand that the Mac version is in Hypercard, which I'm sure you are aware is dead and buried. Were someone to start from scratch to rewrite her program for Mac, I would suggest RealBasic. I used to program in Hypercard (I still have the Danny Goodman books), and RB is like Hypercard grown up. It is a very full-featured RAD tool with a decent IDE, and pretty easy to pick up. I have used it quite a bit myself for several years, and think highly enough of it to pay for the professional subscription out of my own pocket. It has the additional benefit of cross-platform deployment, which I have actually used for some company-internal projects. If you don't have time to do it yourself, you might be able to find a competent programmer who could knock it out fairly quickly (and no, I'm not fishing).

Cheers, Steve Chu

I have the (SuperCard) source code to the old program, but if we were going to do a new version for the Mac we would use the sounds Roberta recorded rather than have the Mac generate speech. (Incidentally, we found that Agnes, rather than the more mellifluous Victoria voice, was more effective for teaching. We're not sure why.) Alas, it would be a great deal of work, and the potential sales do not seem high.

The program runs well under in Windows XP, Windows 98, and Windows ME under Virtual PC on the PowerPC Mac, and in all three of those Windows versions under both Parallels and BootCamp on the Intel based Mac systems. I agree it would be a bit expensive to get either Virtual PC or Parallels plus a version of Windows if her reading program were the only Windows program you intended to run on a Mac. On the other hand, it's probably cheaper than buying a PC for the purpose, and we have had parents do that in desperation. So far as we know, no one has ever finished the program without learning to read English (provided that they speak English in the first place), so perhaps the total expenses are not out of line.

Professor Harry Erwin adds

My experience with Windows XP in Virtual PC is that you really have to push to get the updates to work the first time, but after that they can take care of themselves. You need a broadband connection, several hours to do the update, and (the first time) a willingness to check periodically and deal with any problems. You know, Microsoft owns VPC...

BootCamp has *no* problems at all with updates. I do find it's easier to do the update in BootCamp even if you're using Parallels Desktop for your regular work. (I'm running the beta3 of Parallels that uses the BootCamp partition. You have to authorise the Parallels configuration separately from the BootCamp configuration, but you only need to do it once, and both configurations remain authorised thereafter.)

Harry Erwin

Rich Heimlich is a sounds systems and game design expert. I have known Rich for a long time. I had almost forgotten just how long:

I wanted to remind you of an early call you and I shared with Pierluigi Zappacosta of Logitech. You and I spoke about their so-called first "ergonomic" mouse and it was a joke. It was this half-round circle thing that was just anything but ergonomic. Debbie Fredricks (the PR person for them at the time who would later go on to work for Steve Jobs at NeXT) listened to us and got us into a conference call with Pierluigi. We told him that if they really wanted an ergonomic mouse they should take a piece of clay, lay it under a hand and mimic that shape. He listened, said something like, "Interesting" several times and then said, "Give me six months." At the end of that time Debbie called again and told us they had this new mouse which was entirely ergonomic.

Anyway, I've often wondered if things have gone entirely in the wrong direction ever since then. Obviously they're a huge success and life is good, but I still find their mice to be continually falling short.

I was recently using their G15 gamers mouse and mentioned to you that I had to change its lithium battery at LEAST once every day and often twice. It was quite a habit to have to get into.

Now they've come out with this new MX Revolution mouse and I'm doing my best to give it a full run. I got it as a gift from my son for Christmas so it has to really be bad for me to not use it. I'm starting to wonder.....

They give you this mouse and it's clearly revolutionary in ways. It has two wheels and one can be set to run entirely "free" in that I can spin in and blast through a 50 page Word doc in short order. However, this main wheel has two modes, free rolling and standard where it has the typical roll/detent action. To make that work you push down on the wheel to switch between modes. Only problem is, you lose the middle mouse button now. Just behind the wheel is a button but it's for zooming in documents and its actions can be changed to a wide number of choices EXCEPT acting as a middle mouse button. Here we are, at least 15 years from the battle to get a middle mouse button accepted and they've dropped it in their latest, revolutionary, product.

Then it has a wheel that sits flat in front of your thumb, but it's hardly a wheel. It can be moved forward or back but doesn't spin. It just rocks one way or the other and that's it. It's called the "Document Flip" button and can only be set to flip between applications or zoom mode. That's it.

They have all these buttons and such. When will they figure out that EVERY button, wheel or whatever should be able to be set to whatever function we want? They keep forcing limited sets on us as if we're incapable of knowing how we want to work with a mouse.

I find it interesting that they also went back to the cradle charger for this one. I have 4 of their mice that use the older charger and all of them are dead now because after about 6 months of use the charger can no longer charge the mouse. The connections just fail. Support actually suggests things like licking the contacts before putting it into the charger. Yummy....

Anyway, thought you'd find this interesting.


One day I may try mice designed for gaming. At the moment I tend to use Microsoft Red Eye mice, except that up in the monk's cell where I write fiction I'm using the Microsoft Wireless Keyboard and Mouse. Battery life for those is measured in weeks.

I agree that customizable systems ought to let you customize all their features, but few actually do.