Fun Friday – College Textbook Sticker Shock

I took my daughter Rebecca shopping for her textbooks a few weeks ago at the college bookstore. I walked out stunned with a $300 bill for a soft-cover math book (used) and a soft-cover set of chemistry books (new). And I didn’t even buy any English books yet!

So Michael Granof’s op-ed piece Course Requirement: Extortion in the NY Times hit a nerve with me, and probably every other parent writing those college checks. Granof, a professor at the University of Texas, proposes a “site licence” approach to textbooks based on the projected number of students enrolled, just as a corporation purchases software. Books would be available electronically, or could be purchased in hard-copy form for an additional fee. Instead of being in the paper-pushing business, publishers would become more like software companies focussed on managing contracts for their materials, managing revenue streams, and finding new material and providing updates and revisions. Colleges and professors would be willing to experiment more with classes and new authors, because they wouldn’t be locked-in to the used book market. Textbook authors would find more small markets for their books – it’s all electronic – and could focus on new work and timely revisions for a global economy with deterministic royalties. Libraries and bookstores could invest in “instant book publishing” machines and materials (one machine sampled built an entire book in 15 minutes) and would no longer be risking significant investments in academic inventory (both new and used). And finally, students would find their out-of-pocket expenses for books get more in line with other segments of the book industry.

Hey, as a technologist I’d rather deal with electronic forms of content than hunt for a book on Amazon. As a textbook author, I’d love to spend my time writing new works in operating systems and networking and getting it to students and professors right away rather than worry about whether my older books are being resold and resold until they’re obsolete. And as a parent, I think we’d all like colleges to be in the business of educating our kids, and not in the business of book inventory management.

Of Virtualization and IPOs

VMWare is the little company that could. When it was launched in the late 1990’s, the “smart money” wouldn’t touch it. It wasn’t just that it was open source – a big “loser” badge at that time. It wasn’t just that it was founded in part by a husband-wife team who had established business (Tandem, SGI,…) and academic credentials (UCB, Stanford, MIT,…) – an even bigger “loser” badge that still persists in some dark corners. It’s biggest damnation by old-line investment was that virtualization was a trite idea. I heard this myself frequently from the horse’s mouth, so to speak, during the time I was working with 386BSD and InterProphet, and I pitched a similar idea to Tandem’s CEO back in 1996. It was the next obvious step – at least to a technologist.

Given the tremendous success of this IPO, you might ask “How could they have missed this one?”. Technology, like fashion, goes through phases, and sometimes smart people get so hung up on fashion they miss the trend line change from computer as a calculating device to computer as an organizing device (and vice-versa). If a computer is a calculating device, you want the most cycles. If it’s an organizing device, you want to spend time adding stuff, not managing stuff. A bit processor, like blit-bit processing for graphics, the more concrete, discrete and obvious the instructions executed at the bottom level, the more goodness you get out of the machine. A symbol processor, like parsing language or voice recognition built out of abstraction, is not as deterministic by the nature of the function desired, so capacity to process symbols and the benefit it provides overrides performance.

This happened with virtualization. At that time overfunded Internet companies (remember Egghead, for example, anyone?) and their backers absolutely believed that the most important technical issue was to build a site industriously out of C++ code to maximize performance. They didn’t believe sites based on scripting languages would be powerful or scalable enough for their millions of customers. They underestimated the demand for rapid creation and deployment of new features. Now everyone uses scripting languages like Perl, Python and PHP (we use Python on all our sites for example) – it’s faster and easier. VMWare realized that people allocated servers for containers of scripted sites when performance was impacted, and it didn’t make any difference if it was virtual or real. As reducing power demands becomes a “hot” button topic, virtualization will be increasingly used in datacenter and networking applications.