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

Jerry Pournelle jerryp@jerrypournelle.com
Copyright 2007 Jerry E. Pournelle, Ph.D.

July 16, 2007

The summer season continues. We can begin with an interesting question.

Subject: Outlook?

I am just wondering what it is about Outlook that keeps you using it?

I don't think you have an Exchange server running at home so it can not be the enterprise calendering features. So simply what is it?

I use Thunderbird which for me is just fine. It sends and receives emails, it has a very good spam filter on it, You can create filters, and the latest version has support for both folder, tags, and folders that are custom views of tags!

It doesn't crash on me and frankly works very well. The one problem is that I keep several gigabytes of emails and that can slow it down at times but frankly nothing like what you say Outlook does.

So why do you stay on Outlook?


The one word answer is inertia. Outlook works, although it sometimes drives me crazy, and as we'll see in the column I have yet to get Outlook 2007 working with Vista.

I don't use all of Outlook's features. Indeed, I very much miss the old Franklin Ascend, which used to keep track of my calendar and project management including priorities. Outlook is supposed to do that, but I don't find myself using many of those features; certainly not as religiously as I used Franklin Ascend. (I'd still be using Franklin Ascend except that it died in 2001; it didn't survive Y2K.) The revised Franklin program can't be transferred from one machine to another without a lot of difficulty - at least the one I tried couldn't. Furthermore it would not allow the data files to be transferred by networking. I admit I haven't tried in a couple of years. In any event, I am ready for recommendations for a good calendar, time, appointments, projects, and event management program. I find Outlook's a bit more complicated than I need.

What I do use is Outlook's mail management. I have lots of folders, and rules to sort incoming mail into the folders. That works quite well, and since I get about 100 "real" emails every day and have to answer about half of them, I need something easy to use. Outlook works very well for that. Messages come in plaintext, and if I want I can easily convert to html. Attachments come through properly, and can be dumped or opened as I choose. Message replies are composed in Word with full Autocorrect and spell checking. Autocorrect is important to me; I often type teh instead of the, and autocorrect takes care of that (and similar common mistypes) without my even noticing it. I also like having spell checking.

Outlook works well with FrontPage. I get mail. I can select the mail from the preview window (at least I can if the sender has put the subject in the text message body and not just in the subject line) and "paste special" into FrontPage. This makes dealing with Chaos Manor Mail fairly easy (FrontPage also uses autocorrect and spell checking so replies are easy to compose).

I use Outlook to set up my mailing lists. I keep my subscriber list in Outlook. There's also a courtesy list of people who get my columns by email. Outlook makes all that quite easy.

And finally, Microsoft desktop search understands Outlook. I can find almost any mail from the past few years within seconds. Since I often forget when and where I filed a message, that is also very important.

If I decide to replace Outlook, it will have to be with something that does all that; and so far I haven't decided on what that ought to be. I will also have to have assurance that I can, somehow, convert the Outlook.pst mail files into something that can be indexed and searched. I really need to be able to find things in back mail.

I don't dynamite my life just for kicks, and changing from Outlook would be a major disruption. And first I have to choose what to replace it with. I expect I will get a number of suggestions from readers.

Subject: Use Outlook with Exchange


I'm sure you've heard this suggestion before, and I even have a vague recollection of reading it once on your website. If you want to use Outlook then you should also be using Exchange. Ok, maybe you already are, but if not then my experience has been that Outlook is incredibly slow collecting pop3 email and a hundred times faster via Exchange.

At home, when pop3 email is coming in, I stop what I'm doing go away and leave Outlook for maybe five minutes. But at work I can be typing away in Word, without any performance hit, and not even realise that mail was being delivered until I notice the envelope icon in the system tray. But then the email hasn't actually been delivered, it's still sitting out on an Exchange server.

Very rarely, when I click on an email at work, I see a message like "please wait...your data is being fetched from server..." but usually even very large emails and attachments can be accessed, from the Exchange server, in a matter of seconds.

I know you do the things you do because you are trying to look at things from the same perspective as your readers, most of whom probably don't have Exchange, but given the volume of email you receive and the size of your network the comparisons are never going to be very good.

On the other hand, maybe the real issue is why is Outlook designed to work effectively with Exchange and not with pop3? I guess the home market are less important to Microsoft and they're hoping that hardware and infrastructure upgrades will more than compensate for Outlook's performance before they lose too many customers.

But then I just remembered, you need a solution that would enable you to access your email on the move. Again, I'm sure this has been suggested before, VPN into your home LAN to access mail on Exchange.

Best wishes

Paul Dove

I forwarded that to the advisors conference with the note that I suspected the cure would be worse than the disease. Bob Thompson replied instantly:

No kidding. I've heard that amputation can cure hangnails, too.

If you were running Linux, I'd suggest you use Kmail/Kontact, which is functionally equal to Outlook in nearly every respect, and superior in most.

I certainly never notice any bogging down, and I have many complex filters. I assume that there's a good PIM available for OS X as well.

I'm telling you, Jerry, you have to get away from this Microsoft trap. It used to be that when I read about battered wives, I'd wonder why they just didn't up and leave. Then I realized that it really is human nature to stay with the devil you know. That's the only reason I can think of why so many people put up with this kind of crap from Microsoft when there are much better alternatives.


I don't feel much like a battered wife, but as I noted above, there is considerable inertia in the system. My son Alex added:

Agreed--this is not the cure you want. In particular, he's forgetting that Exchange will require a bunch of bolt-ons (notably anti-spam, anti-virus). Yah, I know you have client-side malware removal, but if you went to self-hosting, all of the spam reduction which happens at your current providers either (1) won't happen because you'll move to the new server or (2) will happen because you are still getting the majority of your e-mail from the other providers, but then you'll have more moving parts (that is: Email goes to Earthlink, forwarded to JerryExchange, then picked up by you).

Yes, you could go to a mail hosting service which uses Exchange, but you'll be swapping one set of headaches for multiple ones. You could use an alternate mail server (we like Kerio MailServer) but I don't think it's worth the hassle.


Nor do I. At this point in my life I want to simplify things. I have been gradually dismantling the artificially large network I built up here to aid in understanding reader networking problems. I keep thinking it's time to update my server system, but then I begin to wonder why I want a server at all; all my desktop systems are powerful and have very large local disks. I need to have them networked but they don't need to operate through any server system.

But there's still a lot of inertia in the Chaos Manor system.

I get a lot of press releases. I try to read most of them, but that's about as far as it goes. Once in a while I get one I pay more attention to. I had a look at the attached ebook, then I sent this to the advisors, with my usual question.

Is this interesting?

To: Movers, Shakers, and Other People We Like :

Subject: "Take Control of Running Windows on a Mac" for you

Hi everyone,

We've just released a fairly major update to Joe Kissell's "Take Control of Running Windows on a Mac" that includes the very latest information. If you follow this topic on the Mac at all, you'll know that this is quite an achievement, what with recent releases of the Boot Camp 1.3 beta, Parallels Desktop 3.0, the first release candidate of VMware Fusion, and the free public beta of VirtualBox for Mac OS X. Joe was adding new details up to just a few days ago to keep up.

Since anyone in your position should have the latest details about Apple's Boot Camp software and virtualization products, I've attached a copy of the ebook to this message. I've also included the press release below. Of course, if you'd prefer that I stop sending you copies, just let me know.

Also drop me a line if you have any questions or interview requests, if you'd like to publish an excerpt of the ebook, or if you just want to chat.

cheers... -Adam


Freshly updated Take Control ebook provides the latest advice, tips, and steps for successfully running Windows on an Intel-based Mac.

Ithaca, NY USA (July 12, 2007) -- People who want to run Windows XP or Vista on a Mac have plenty of options, including Apple's Boot Camp, Parallels Desktop 3.0, newcomer VMware Fusion, and the free VirtualBox. But which to pick? The completely up-to-date "Take Control of Running Windows on a Mac" provides expert advice and step-by-step instructions from author Joe Kissell on choosing the best technique and software for getting Windows running on a Mac. The ebook also includes real-world advice about installing Windows, dealing with tricky peripherals, sharing files between Windows and Mac OS X, backing up a Windows installation, avoiding Windows malware, and more.

"Take Control of Running Windows on Mac" is available for $10 at http://www.takecontrolbooks.com/windows-on-mac.html, and it includes coupons worth $10 off Parallels Desktop 3 and $5 off any order at Small Dog Electronics.

Take Control editor in chief Tonya Engst said, "This ebook has been a huge effort, given the dramatic changes in the topic over the last few weeks, but Joe has done a great job of incorporating new information up to the very last minute. No other book-length title has the latest details about virtualization on the Mac, and I'm thrilled that we've been able to make it available to readers so quickly."

Book Details:
"Take Control of Running Windows on a Mac" by Joe Kissell
PDF format, 148 pages, free 29-page sample available Publication date: July 11, 2007 Ebook
Price: $10

Peter Glaskowsky replied:

Yes, for two reasons:

Adam C. Engst has been an important figure in Mac publishing for a long, long time now. Pretty much anything he thinks is important about the Mac ecosystem probably is.

In this particular case, I read the sections of this book that cover the things I've just been doing-- getting Vista running in Boot Camp and sharing the Boot Camp partition with Parallels, so now I have only one Windows install on the hard disk (or will, anyway, once I move the old Parallels XP install to a backup disk).

The book seems to be correct in all the particulars I can evaluate. I would have preferred to see more detail about potential problems and other extraordinary situations, but most people will find the book to be entirely sufficient.

So yeah, this book is probably worth mentioning to your readers.

. png

I have several letters inquiring about registry cleaners. I also get press releases advertising them. I ignore most of them, but I have some experience with Wise:

Company: Wise Cleaner
Title: CEO/Founder
E-mail: support@wisecleaner.com

Achieve Peak PC Performance with a Freeware Registry Cleaner

Wise Cleaner today announces the availability of Wise Registry Cleaner 2.8, a freeware tool that enables you to clean Windows Registry safely and quickly.

Computer users often install new software for evaluation and uninstall it if it doesn't meet their expectations. Each time you install an application it writes a few entries in Windows Registry, when the software is uninstalled the entries often remain in registry. Over the years, your registry accumulates a lot of entries from the software you no longer use, and this has a negative impact on the system: applications slow down and run with errors, you are often kept waiting for the program to launch. To achieve top performance again, you don't have to buy new hardware. Simply clean your registry and the computer will run like a new one!

Wise Registry Cleaner will scan registry for incorrect and obsolete entries and let you fix found issues to make your computer run at the peak speed it's capable of. The scan is quick thanks to the enhanced scan engine, and after a short while you can see a list of issues that need to be repaired.

You can view each issue individually, bring up its detailed description to help you figure out how to deal with it best and repair it or leave unchanged.

For your peace of mind, the program will automatically back up all changes done to registry, so that if there is any wrong change, you can undo it easily with a click. If you are a non-techie user and know little about Windows Registry, you can rely on the program's Artificial Intelligence and let it repair issues automatically.

The program has a friendly interface that you can familiarize yourself with in the first few minutes of using the program. Even people who are not exposed to computer technologies can get started with the software without reading its user manual.

New to version 2.8 is enhanced security, improved multi-language support, an ability to report when some entries cannot be removed and some other features.

Pricing and Availability

Wise Registry Cleaner 2.8 runs on Microsoft Windows 9x, NT, 2000, XP, 2003, Vista including x64 version. Additional information on the product, its manual, as well as the freeware downloadable version is available from www.wisecleaner.com.

I asked for comments on both this and registry cleaners in general:

Has anyone had one that really worked? Mine seems to clean like an SOS pad cleans a car--paint and all. Sooner or later you find out it wiped out something you needed 3 weeks after it did it.

Rich Heimlich

Which pretty well sums up my experience with registry cleaners. Either they do very little, or they muck up something better left alone. Eric Pobirs adds

They can be of value but the system in question needs to have spent a lot of time associating with software from the wrong side of the tracks.

Other than that the culprit is accumulation of bad entries over time but even that requires some changes to the system being a regular occurrence, like installation and removal of games. Unlike the 9x days, an XP box isn't nearly as subject to build-up of registry gunk from running the same reasonably behaved software for years on end.

But some of those really bad apps can do a lot of damage all by themselves. Some of the changes in Vista specifically locks out some of the worst behavior and is where many of the incompatibilities occur, such as with Intuit apps prior to the 2007 editions. MS made this part of the XP logo compliance specs almost seven years ago but many companies ignored this until Vista raised it from a request to an enforced requirement.


I downloaded the Wise cleaner, and it does make backups before it begins - something you should always do before hacking into the registry. It found a few things it didn't like in my registry, but I can't say I have noticed much difference in system performance since using it.

Eric notes that the systems that really need a registry cleaner are those that have been sent to bad web sites or have been running bad applications. The problem here is that a registry cleaner may not be drastic enough for some of those bad apps. It's unlikely to deal with sophisticated malware: some of that stuff, including root kits, requires nuking the system from orbit, scrubbing down to bare metal, and reinstalling everything.

I've played with registry cleaners for a decade, and back in the Windows 9X days hacking the registry to get rid of unwanted tag ends could be important, but I haven't felt the need to use one on an XP system.

Last week we talked about the Xbox 360 problems. It generated discussion among the advisors. We have a lot of discussions, some pretty heated, in my advisors conference. This one began with:

Subject: Xbox 360 problem universal

I see that Microsoft admitted today that all 11.6 million Xbox 360 units that have been shipped to date are defective. No recall is planned.

The high figure announced by Microsoft for extending the warranty was $1.15 billion, which comes to "only" about $100 per system shipped. Counting fulfillment costs, that implies that the problem is relatively small and easily fixed. I wonder if that's true. If not, the true cost could easily be twice or three times what Microsoft just wrote off.

I've never even seen an Xbox 360, but I understand things are pretty tighly integrated. I'm not an EE or materials scientist, although as a former ham operator and holder of a 1st-class radiotelephone license I do have some electronics background, along with some long-ago courses in materials science.

If someone put a gun to my head and forced me to speculate, I'd guess that it's a mainboard problem, and that the root cause is faulty soldering. The Chinese companies that produce the Xbox 360 probably tried to save literally a penny or two per unit by using cheap solder that was outside specs (if indeed Microsoft even spec'd things properly.)

It would be interesting to examine some failed units under the microscope. I'll bet you'd find two things: tin and/or indium whiskers and failed solder joints from thermal cycling. If I'm right, this is reminiscent of the problem with cheap capacitors that cost motherboard manufacturers and OEMs billions of dollars.

Robert Bruce Thompson

I had my doubts about his theory on the cause of the problem, but it did seem interesting. Eric Pobirs came in with:

Very doubtful. The leading suspect for the rate of Xbox 360 failures is heat, a belief bolstered by the fact owners who opened up recently repaired units found additional heatsinks now in place beyond those originally included in the design.

Microsoft and Sony tried to get as much as possible from 90nm production for boxes that have to be sold at a loss to reach the large numbers needed.

Sony has had far less reliability issues thus far but at the penalty of a more expensive product after the differences in optical drive and hard capacity are factored out. (As it is, evidence suggests the original multiple Cell design for the PS3 was unviably expensive and scrapped, and Nvidia brought in fairly late to supply a dedicated GPU.) Nintendo has chosen to get off the high-tech merry-go-round entirely, opting instead to base their Wii on a fairly minor upgrade and die shrink of their previous generation product's chipset. Innovative user input controls are the driving feature of the Wii, along with Nintendo's treasured franchise library.

A 100% Xbox 360 defect rate is extremely improbable. The highest credible figure, from a major retail chain executive, was 33% and even that is likely overblown. The fact is that 100s of thousands of machines from the launch period have never failed. The numbers don't need to be very high to cause concern. Sony was hit with a costly class action suit over PS2 failures but nobody ever claimed number far into the double digits with a straight face. Microsoft is smart enough to know it is better to take action before action is taken against them. That $1.15 billion is a worst case scenario, which accounting laws require a company to state when taking such a charge. The actual cost will be less, potentially far, far less.

Microsoft didn't test the design sufficiently before going to market but the cost could have been far higher, both in missing that first Xmas season with no other next-gen competition and in letting the failure rate become a matter for the courts.

Considering what happened to Toshiba in the bizarre floppy controller case, where no real-life occurrence of the defect could be documented, keeping it out of court was a very wise decision.


Which I think pretty well rules out the soldering defect hypothesis. Dan Spisak added:

I have a launch XBOX 360 (not given to me by MS, paid for in my own money) and it hasn't exhibited any problems. Another friend of mine also got her 360 at the same time at the launch event and has had to have hers replaced once under warranty. I don't know about anyone else with one personally.


Of course Rich Heimlich has an XBOX 360.

Ours is fine as of now. Hope it stays that way. We got ours fresh from the launch and I've worried about it a bit but so far....


Peter Glaskowsky comments:

There are no Chinese companies producing the Xbox 360, unless you count Taiwan as part of China.

The three manufacturers (as far as I know) are:

Celestica, headquartered in Canada with operations in 16 countries, including 12 facilities in North America.

Flextronics, headquartered in Singapore with a predominately American management team.

Wistron, a Taiwanese company with manufacturing facilities in four countries.

I think this broad base of manufacturing operations makes your hypothesis about soldering errors on 100% of all units sold completely impossible.

. png

At this point the nature of the discussion changed: precisely what did Robbie Bach say, and more important, what did it mean? Eric Pobirs began:

I haven't seen any quotes from Bach indicating he thought the problem was in potentially 100% of the units. Probably because it isn't, considering units produced in recent months are a few revisions different from those at launch.

OTOH, Bach has been quoted recently as stating his belief that the cost of the warranty extension doesn't change their expectation to be in profits on the Xbox 360 in the next year. Establishing consumer confidence at this time is critical for the September 25th release of Halo 3, which will likely have an attach rate of over 50%. Something only a handful of games ever achieve. If it manages anything like the popularity of its predecessor that one title could gross over $300 million by Xmas.


I went looking for the original interview story, but first I had another look at the story Bob Thompson quoted to begin this:


It was pretty straightforward:

Microsoft facing US$1.15bn Xbox 360 repair bill

By Tom Sanders, 9 July 2007 07:27 AEST Consumer Electronics

Software giant admits there are 11.6 million faulty consoles, which it will have to fix.

Microsoft has admitted that every one of the 11.6 million Xbox 360 consoles sold in the past 19 months suffers from a design flaw that could cause the device to fail.

Eric observes that:

What you've presented is the writer of the article stating that he believes MS has in effect made such a statement. What you did not offer is the MS exec or anybody else from MS saying such.

One thing that has to be remembered here is the internet effect. This is where a handful of disgruntled consumers complaining in every forum they can find having far louder voice than satisfied users of vastly larger numbers.

A guy with a broken console is going to talk about that while the many others with working consoles are going to talk about the games they've been playing rather than the continued operation of the console itself.


That sent me off to find the original interview. I found:

David Hilal (Friedman Billings Ramsey): A question on how many units have been affected out of the 11.6 million and is there a problem that's completely behind it?

* Robbie Bach: Yeah, so we're not going to discuss, nor have we historically discussed, return rates or specific numbers of units. Suffice it to say that with a billion dollar charge and the focus we're putting on this that it's a meaningful number. It's one that we take very seriously and one that clearly has our attention. In terms of going forward, we do feel like we understand the issues and have made the changes needed to dramatically reduce this problem going forward and that we think we have our hands around it at the engineering level, which is the important thing for us clearly to do. And in fact, as Chris inferred, if you look at our fiscal year '08 expectations those are on track with what we said in the past and that's because we feel comfortable with where we are on the engineering side.

* Alan Cook (Merrill Lynch): This product's been out for a little over 18 months. Can you tell us when the problem became apparent? Is it something that's just cropped up over the past few months? Or has it been around for the full 18 months?

* Robbie Bach: Well, for a little over the first year, this problem - this set of issues - wasn't visible at all. It's the type of problem that doesn't happen because you turn on the 360. Sometimes you have hardware problems the first time you turn something on and you have a problem right away. These aren't those types of problems. **So for the first year-plus it was something that wasn't, frankly, really on our radar screen.** But in the last couple of months, we started to see significant increases in repair requests, significant call volume, and significant attention from people.

And so we geared up to respond to that appropriately. It is one of those challenges in terms of figuring how you can test for things that happen a year to 18 months into the life cycle of a product; it's the type of testing that we do today and we're gonna do obviously more of that going forward.

**emphasis added**

It is a fair inference that this is a design flaw, which means that EVERY DAMN ONE of the systems is vulnerable. It probably depends on how much use they get. Some of those things go on a shelf and come off on Independence Day and not otherwise, and it would be needlessly expensive to recall them all. They wait for them to fail, but they have extended the warranty, which seems like a good thing to do.

It may be that the headline

Software giant admits there are 11.6 million faulty consoles, which it will have to fix.

Microsoft has admitted that every one of the 11.6 million Xbox 360 consoles sold in the past 19 months suffers from a design flaw that could cause the device to fail.

Is an overstatement, but it's defensible given what Bach said.

Jerry Pournelle

Peter Glaskowsky adds:

Some quick facts:

Bach and the other Microsoft participants DID NOT SAY that all units have defects.

In fact, during the Q&A session, Microsoft specifically declined to give any estimates of the number of units that have been or might be affected.

Robbie Bach did describe the source of the problem as "a Microsoft design challenge" which in context I took to mean a "design flaw", the phrase that has been widely used.

Design flaws, of course, do not always result in product defects. If a design results in only 99.9% reliability when the goal was 99.99% reliability, I-- as an engineer-- would call that a design flaw.

In this case, of course, it's more likely that Microsoft fell far short of its target-- more like 90% instead of 99.99%. Those numbers are not from Microsoft, I'm just using them to make the point. The actual failure rate is probably at least 100 times higher than Microsoft originally intended.

Bach also said, specifically, that Microsoft's component suppliers and manufacturing partners were NOT at fault in this case.

The CRM writer is obviously not an engineer and had no legitimate basis for saying what he said.

The final key point is that Bach said Microsoft has solved the design issue.

. png

I put it to you that a difference of 10% may be more meaningful to an engineer like Glaskowsky than to Bob Thompson, or for that matter, to me. I infer from what he said that it's a design flaw, and certainly a design flaw has the potential to affect 100% of the units. Of course many -- possibly as many as half -- won't be used under conditions that trigger the defect. Most of the defective seat belt anchors that caused so many automobile recalls did not result in actual breakage. In fact very few did. But they were all recalled.

This isn't a critical thing like an auto seat belt and it makes sense to take care of things as they break, not do a recall; but I certainly can't fault anyone for inferring from Bach's statements that all, 100%, of the units are at risk. Of course not all will fail, but it's still a fair inference.

The difference between design flaw and product defect is worth further discussion, but this is long enough for now. Long as it has been, I think it has been instructive. It was for me.