Previous Issue | Search TidBITS | TidBITS Home Page | Next Issue
What would you like Apple to learn from last week's iTunes installer debacle? Adam looks at some take-home lessons for Apple and passes along some advice for the rest of us. Dan Kohn's series of essays on the future of content in a digital world continues with a look at ways of financing pure public goods. And in the news, we cover updates to Microsoft Outlook Express 5.0.3, Adobe Illustrator 10, ConceptDraw 1.71, DAVE 3.1, and IPNetRouter 1.6.2.
Copyright 2001 TidBITS Electronic Publishing. All rights reserved.
Information: <firstname.lastname@example.org> Comments: <email@example.com>
This issue of TidBITS sponsored in part by:
READERS LIKE YOU! You can help support TidBITS via our voluntary
contribution program. Special thanks this week to Wayne Marsh,
Robert Estler, and Jerry Kerrisk for their generous support!
APS Tech -- 800/395-5871 -- <firstname.lastname@example.org>
Burn a full CD in less than five minutes with the APS CD-RW
16x10x40 FireWire Plus. FireWire and USB ports let you easily
connect to all recent Macs. Order at: <http://www.apstech.com/>
WinStar Northwest Nexus. Visit us at <http://www.nwnexus.com/>.
Internet business solutions throughout the Pacific Northwest.
Small Dog Electronics: Apple 17" CRT Studio Display: $319!
New iMac G3/500 28/20 GB/CD-RW/56k Flower or Dalmatian: $849!
PowerBook G4/400 128/10 GB/DVD Titanium Factory Refurb: $1,679
Norton SystemWorks: $77.77 <http://smalldog.com/> 802/496-7171
Maximize Your Internet Connection! Ever wonder if you can get
more speed out of your Internet connection? Get IPNetTuner and
squeeze every possible bps from any Internet connection. Only
$25 from Sustainable Softworks! <http://www.sustworks.com/tb/>
NEED TRAINING? Call MacAcademy! Now in our 15th year of business
we offer the highest quality, lowest priced, most effective
training for today's computer user. Visit us at:
<http://www.macacademy.com/tidbits.html> or call 800/527-1914
Bare Bones Software BBEdit 6.5 -- New version adds CSS markup
and syntax coloring, even more powerful grep engine, integrated
Unix shell features on Mac OS X, and much more. Buy, upgrade,
or try the demo at our Web site: <http://www.barebones.com/>
Outlook Express 5.0.3 Fixes Access Problems -- Microsoft has released Outlook Express 5.0.3, a maintenance update of the free email and news client. The update restores the capability to access MSN Hotmail accounts, which was disrupted recently when Microsoft modified its Passport servers. In addition, Outlook Express 5.0.3 adds support for accessing MSN Internet accounts and offers enhanced SMTP AUTH support for authenticated transmission of outgoing messages. The update is a free 9.1 MB download. [JLC]
Adobe Releases Illustrator 10 -- Adobe is now shipping Illustrator 10, the first of its graphics applications to add Mac OS X compatibility. The new version adds several features for producing Web content, including the capability to slice vector images, enhanced Macromedia Flash support, and capabilities that take advantage of Adobe's Scalable Vector Graphics (SVG) format. Other improvements include liquefy tools for warping images, symbol tools for managing repeated elements, a Magic Wand tool for selecting similar objects, and a host of interoperability options for using Illustrator with other Adobe applications. Illustrator 10 requires a PowerPC G3 or G4-based Macintosh running Mac OS 9.1 or later (including Mac OS X 10.1) and 128 MB of RAM. The program costs $400, with $150 and $250 upgrades available for owners of previous versions of Illustrator or competing products, respectively. In Adobe's announcement, President and CEO Bruce Chizen said that the rest of the company's titles would be released for Mac OS X within the next six months. [JLC]
ConceptDraw 1.71 Released -- CS Odessa has updated the Standard and Professional versions of ConceptDraw, its software for creating diagrams and flowcharts (see "Make the Connection with ConceptDraw" in TidBITS-553 and "CD-Odessa Takes ConceptDraw Professional" in TidBITS-597). Changes in this release include support for PNG graphics, improved EPS export and data exchange with Adobe Illustrator and Photoshop, and a few bug fixes and optimizations under Mac OS X. The Standard version also boasts a streamlined user interface and more shortcuts and menu commands. The ConceptDraw 1.71 update is free to registered users, and is a 4.5 MB (Standard) or 4 MB (Professional) download. [JLC]
IPNetRouter 1.6.2 and Continued Care Available -- Sustainable Softworks has released IPNetRouter 1.6.2, the latest version of the company's popular software router for sharing a single Internet connection among multiple computers. Improvements include a larger filter table for the built-in firewall, the capability to turn the company's IPNetSentry firewall software on or off from within IPNetRouter, some cosmetic bug fixes, and developer extensions to the DHCP server functionality. The update is free to registered users and is a 1.5 MB download.
Separately, Sustainable Softworks announced Continued Care, an optional $25 per year subscription support plan for customers who have not paid for a product or upgrade within the last year (those who have are entitled to free support for one year). The Continued Care program gives subscribers support queue priority for all Sustainable Softworks products (including the just-released entry-level Mac OS X Internet sharing software, gNAT), direct contact with developers, and both email and telephone support. [ACE]
DAVE 3.1 Adds Mac OS X Support -- Thursby Software Systems announced the availability of DAVE 3.1, the newest version of the company's utility for sharing a Mac on a PC network. In addition to bidirectional file sharing between Macs and Windows-based PCs, DAVE 3.1 adds the capability for Windows users to access Mac machines and printers (both inkjet and PostScript) under Mac OS X 10.1. Under Mac OS 8.6 and later, only PostScript printers are accessible. Other features include large file and file name support under Mac OS 9 and Mac OS X 10.1, and automatic workgroup detection. DAVE 3.1 is a free update for anyone who purchased DAVE 2.5 during the 2001 calendar year; older copies and previous versions can be upgraded for $90, while a new version costs $150. [JLC]
by Adam C. Engst <email@example.com>
The weekend of 03-Nov-01 was a bad one for Apple and some early users of iTunes 2. After releasing the new version late Friday night, Apple hastily pulled the Mac OS X installer Saturday morning due to a problem where, in some situations involving multiple volumes named in specific ways, the installer could delete a large number of files. Needless to say, this is a bad thing, and there have been reports of Apple quietly offering to buy file recovery software or even pay for DriveSavers recovery of affected hard disks. A revised installer, with the designation iTunes 2.0.1, was released before the end of the weekend.
The specifics of how this happened have been discussed at length in TidBITS Talk and similar forums, but roughly speaking, the installer Apple used to install iTunes in Mac OS X apparently relied on a shell script that assumed the previous version of iTunes would be in the Applications folder. Since everyone's disks have different names, the script figured out the name of the disk, appended the path to the iTunes application, and then deleted all the files in the iTunes folder. Unfortunately, the script didn't take into account the fact that people might put spaces in their disk names, particularly that they could put spaces at the beginning of the disk name. Since the space separates arguments in Unix commands, a command that would delete a single file is suddenly broken in the middle, transforming it into a command that can delete an entire disk. The problem can be avoided in Unix merely by enclosing the command in quotes, but that didn't happen initially.
It's easy to blame Apple for sloppy work and to bemoan the loss of data by innocent users. But I think this event points out a number of deeper issues that underlie the entire move to Mac OS X, showing precisely where Apple needs to work at a variety of levels, including the cross-pollination of Unix and Macintosh knowledge, the use of appropriate installation technology, and the crying need for backup support.
Mac Plus Unix -- Much has been made of the schism between Macintosh and Unix, with Apple saying that you won't need to know any Unix to use Mac OS X and Unix geeks gleeful about getting an operating system that can run both Unix software and mainstream productivity applications. What this installer debacle shows is that Mac OS X developers and experts alike will have to be fluent and comfortable in both the Macintosh and Unix worlds. Look at the mistake that was made in the Mac OS X iTunes installer to see why.
From one point of view, the person who built that installer was clueless about Unix. Anyone with any real Unix experience knows that you have to quote Unix pathnames that contain spaces, but that's something a Macintosh user would never even consider as a concern. However, it's equally easy to surmise that the person responsible for the installer had no idea that normal Mac users are perfectly capable of naming hard disks with spaces - even leading spaces - not to mention untold other troublesome characters like leading hyphens. A Unix person would never consider doing such a thing.
It doesn't really matter which possibility was true in this case - my point is merely that without widespread knowledge of both operating systems, and beyond the operating systems to the usage conventions and opportunities afforded by each, mistakes like this will continue to happen.
Installation Automation -- I may be giving too much emphasis to the lack of knowledge on the part of the person who built the installer - it's entirely possible this person is the pinnacle of both Macintosh and Unix knowledge and made a simple, human mistake. It happens to all of us, and it's certainly easy to imagine the omission of a few quote characters. This raises two points.
Why did Apple choose to use its own installer with a hand-written shell script, where there's no pre-tested code and interface that works to avoid simple human errors? Keep in mind that this is the installer that already has several black marks against it (one relating to passwords with unexpected characters, another with blown permissions).
What's especially ironic here is that Apple used MindVision's Installer VISE for the Mac OS 9 iTunes installer, which would have lowered the effort and costs of building a Mac OS X version of the installer. Installer VISE could even have distributed both the Mac OS 9 and Mac OS X versions as a single installer file, installing appropriately depending on the operating system in use. It's certainly conceivable that someone could have made a similar mistake with Installer VISE, since Installer VISE can pass installation paths to Unix shell scripts, but there wouldn't be any need to write those shell scripts. One way or another, Installer VISE is a long-standing, heavily used installer that's been tested both by MindVision and a veritable army of Macintosh developers while building installers for thousands of products. As all Mac software should, Installer VISE goes out of its way to make sure the right thing happens. The same is certainly true of Aladdin's InstallerMaker, which I used back when I was creating installers for my books.
I'm sure there are numerous possible reasons Apple chose its own installer, including pride, the dreaded "Not Invented Here" syndrome, the fact that Installer VISE comes from the Mac world instead of the Unix world, and even the laudatory goal of using one's own tools - "eating your own dog food," as it's called in the industry. The specific reason doesn't particularly matter; what matters is that Apple learns to focus primarily on what's important for customers. If it's too easy to cause data loss using the Apple installer, that's a problem. It's possible that the just-released Installer Update 1.0 (available last Thursday, via Software Update) will help, since it "delivers improved support for installing software updates and is required for any future Mac OS X updates." Unfortunately, we have no way of knowing, since the phrase I quoted above is the sum total of what Apple's saying about the improvements. And, not that it really matters for this discussion, why didn't Apple release iTunes 2 via Software Update?
Where's Your Backup? Perhaps the most commonly uttered platitude in the computer industry is, "Make sure you have a backup, then..." We've said it innumerable times in TidBITS, and it's standard advice when installing new software. Of course, no one should be making a backup specifically to prepare for installing software - you should have a backup strategy that ensures you're adequately protected at any given time.
The big problem is that you still cannot back up a Macintosh running Mac OS X easily, reliably, and automatically in such a fashion that you can restore the entire machine to working order in the time it takes to read the files from your backup media. Yes, there are some ways of copying important files to another hard disk or machine over the Internet. But there is no way to implement a real backup strategy right now, which involves automated backups that run on a regular schedule, copy only changed files, preserve all permissions and file attributes, and keep different versions of the files. Backup software running in Mac OS 9 or under Classic cannot back up all Mac OS X files such that they can be restored properly, and even the public beta of Dantz Development's Retrospect Client for Mac OS X has troubles under Mac OS X 10.1.
Apple simply hasn't devoted the resources necessary to making it possible to perform backups because backups aren't sexy, don't sell boxes, and imply that data loss is likely. Realistically, though, data loss isn't a question of if, it's a question of when - at this point, the main reason I refuse to install Mac OS X 10.1 on my primary Mac is that I can't back it up acceptably. Those who lost data because of the poorly written iTunes installer would have been annoyed at the loss even if they'd had backups, but the damage wouldn't have been nearly as great. These people didn't have backups (or at least good backups), and this time it's not just a case of the user being lazy or tempting fate. This time Apple deserves the lion's share of the blame for creating an operating system that can't be backed up and restored reliably many months after the initial release. For this reason alone, Mac OS X cannot be considered acceptable for serious use in many situations.
Lessons -- In the end, it's worth remembering that getting everything right all the time is near-impossible, and Apple at least reacted quickly to the problem, pulling the Mac OS X iTunes installer and posting a warning. What's most important is that Apple learns from this mistake so something similar doesn't happen again. Can you imagine the fallout if this particular problem had been on all the Mac OS X 10.1 CD-ROMs?
For Apple, then, I'd recommend the following:
Encourage the cross-pollination of knowledge between Mac and Unix experts, both inside and outside of Apple.
Use an installer technology that works to reduce the likelihood that a simple human error could cause data loss and increase testing on any installer that can delete files.
Make it possible to perform real backups of Mac OS X machines as soon as possible.
For the rest of us, I recommend waiting a few days before installing any updates. Others will always be more brave (or foolhardy), and the more patient among us should learn from their experiences.
by Dan Kohn
"If consultants had been hired to evaluate the market for printing a decade or two after its invention, they would have concluded that the new technology was vastly overrated. Scribes were already producing the important books efficiently, and the new printers produced mainly the same old texts, such as the Bible, which were readily available to the tiny minority who were literate."- Ithiel de Sola Pool
Who will pay for something "free?" It seems obvious that content creation has to be funded somehow, since the next Jurassic Park won't be developed during some teenager's free afternoons. Economist Arnold Kling says that "The central paradox of our times is that information wants to be free but people need to get paid."
Pure public goods have been funded by various means for the last several hundred years, so there should hopefully be some insights by now into how the makers of pure public goods can be compensated for creating their works. For if these options aren't realistic, it is unclear why authors such as this one will continue as ink-stained wretches (well yes, this essay was typed directly into the computer, but you get the idea).
There are four basic ways to fund a public good: the government, micropatronage, funding from non-profit organizations or corporate philanthropy, and the sale of atoms associated with the bits. These are in no way exclusive - it's entirely likely that any given public good may receive funding from more than one of these sources.
Government Support -- Although few people like the idea of the government getting to decide which artists and writers "deserve" to be funded, government support has been the traditional manner to fund public goods. Given the political histrionics surrounding the U.S.'s National Endowment of the Arts, it seems unlikely that this model will scale to fund a much larger supply of content. Neither the public at large nor Jesse Helms himself want to live in a world where the Senator decides which situation comedy or hip-hop band is most deserving of funding.
However, there are alternatives that could enable much larger government funding of content creation without involving the government in picking winners and losers. These measures would be based directly on popularity, so that creators of popular content would be compensated for doing so. For instance, a commercial service called MediaMetrix currently calculates the most popular Web sites, just as Nielsen's delivers ratings on the most popular TV shows. A new government program could offer grants to the authors of the top 1,000 most popular Web sites, MP3s, movies, and so on, encouraging innovation not just in the creation of new content but also in its successful promotion.
Micropatronage -- Don't confuse micropatronage with micropayments, which make no sense for pure public goods. Specifically, if anyone who pays $1 for a New York Times article can redistribute that article endlessly, why would anyone else pay the dollar? Plus, people are mistrustful of the ways micropayment systems have been proposed thus far, due to the fear of fainting spells when reviewing the bill at the end of the month.
Micropatronage entails a return to the content creation system of the 15th century, namely art patronage. The downsides are that artists are open to influence from their patrons (though arguably less than they are by their publishers today) and that content creators need to be matched up with patrons. The advantage is that with the near-zero cost of information distribution, finding patrons becomes simpler, as does having one artist supported by numerous patrons. Thus, rather than the Medicis funding Michelangelo's works, an artist such as Aimee Mann with a small, passionate following could probably find 1,000 individuals willing to donate $100 a year or more. (The main difference between modern micropatrons and the Medicis is the dramatically smaller likelihood of going from being a patron of the arts to becoming the Pope.)
Is there any model of micropatronage in action today? Yes, public broadcasting. Fund drives for PBS and NPR run into a basic difficulty that they are trying to raise money for a pure public good that is available to everyone. But large numbers of individuals still seem to find the money to donate, even though they could not be excluded from future content for not doing so. (Or phrased differently, their guilt from enjoying PBS or NPR outweighs the economic certainty that no single person's contributions will have any effect on whether they get to continue enjoying this nonexcludable good.)
Mickey Kaus's personal news Web site recently declared profitability (although this is misleading given that he is not charging for his own time), where he is relying on an Amazon-based system to enable readers to become micropatrons. Even TidBITS uses micropatronage, with over 700 contributors to date.
More generally, almost all charitable giving today falls under this model, in which the charitable services (e.g., the Salvation Army or your school's alumni association) are equally available to you whether you contribute or not, but you still choose to do so. In situations in which there is no downside for any given individual to be a freeloader, it is amazing that so many charitable services continue to survive on donations alone.
Non-profit Foundations and Corporate Philanthropy -- Another way that pure public goods such as medical research are funded today is through non-profit organizations such as the American Cancer Society. These raise money for a single larger goal and then distribute it to the uses they believe are worthiest.
This approach could easily apply to content as well: imagine a National Country & Western Foundation, or an American Society for Horror Flicks. That is, some potential micropatrons (especially companies) may wish to fund a genre as a whole, and have the experts employed by non-profits decide which established and up-and-coming artists are most deserving of their funds.
The Web magazine Slate's current financing seems to fall into a similar category, whereby a (very) for-profit corporation, Microsoft, apparently feels that the respect and gratitude it garners by funding a high-quality magazine is worth the $20+ million a year that it spends to operate Slate.
In fact, nearly all influential news and opinion magazines are financed by individuals and/or companies that appreciate the respect that seems to rub-off from financing top rate content. Examples include The New Yorker (Si Newhouse's Advance Publications), The Weekly Standard (Rupert Murdoch's News Corp.), and The New Republic (Marty Peretz). They certainly are not run on any generally understood financial principles, since they all lose money.
As content becomes less excludable (meaning mainly that people are reading or listening via digital devices rather than through paper and CDs), many new such foundations will likely be needed.
Sell Atoms Associated with Certain Bits -- Finally, one system that is beginning to work for funding a public good is to charge for physical items associated with given content. That is, even though bits are becoming impossible to charge for, many people will be willing to pay for the atoms associated with those bits. The most obvious are concert t-shirts, or micropatron plaques with a signed thank you from the content creator. Since (all but open-air) concerts are by definition excludable, fans will continue to pay real money to attend them. In the case of concerts, the atoms that fans are paying to be in proximity to are those of the artist herself.
Are there really enough funds available from these four methods - even combined - to pay for the next Madonna CD or an episode of The West Wing? Only time will tell. But, just as technology is nearly eliminating the cost of distributing information, so too is it drastically reducing the cost of creating content. The dinosaurs that took 50 people and dozens of supercomputers to create in the original Jurassic Park may soon be reproduced by a 16-year-old working after school on her home PC. The symphony that Mozart had to assemble an orchestra to hear can now be authored (and iterated, and iterated again) with a free music program in an afternoon (presuming the talent), with no musicians to train or pay. Of course, even these teenagers will need to pursue some combination of these funding mechanisms if they want to have their art be more than a hobby.
And those four options are it. Although I (and numerous hungry artists) will continue looking for new funding innovations, I can't currently envision many other ways that pure public good content could be funded. Which leads to the question addressed in the next essay, of whether all of this is fair.
[Dan Kohn is a General Partner with Skymoon Ventures. His writings are announced through <firstname.lastname@example.org> and can be discussed through <email@example.com>.]
Non-profit, non-commercial publications and Web sites may reprint or link to articles if full credit is given. Others please contact us. We do not guarantee accuracy of articles. Caveat lector. Publication, product, and company names may be registered trademarks of their companies. TidBITS ISSN 1090-7017.
Previous Issue | Search TidBITS | TidBITS Home Page | Next Issue