OpenSolaris, VMWare, Sun Developer Network

I'm wanting to play with OpenSolaris... Solaris seems technically cool, I have a lot of respect for the engineering prowess of Sun, and ZFS and DTrace have gotten a lot of publicity and seem cool.

To a certain extent experimenting with OpenSolaris runs contrary to my overall feelings on operating systems at this point, which is that the important thing is a convenient, well-thought-out packaging and package-update mechanism, with the largest possible number of things packaged for it, and ideally available from one or a few trusted sources. This is why I like Debian so much.

Another issue is whether such an experiment is the right way to handle the fact that there are always too many things to learn. Software is all about lots and lots of precise and largely arbitrary details, resulting in a stiff learning curve when trying to use something new. (And, correspondingly, great gains in power that come only from deep familiarity with a system.) It is a key professional skill to manage this learning effectively: on the one hand, it is easy to get too deeply entrenched in what you already know how to use, and miss out on opportunities arising elsewhere; on the other, it is easy to obsess over always learning new things from the beginning, when gaining further skill with things you already use and know a little about would be more valuable.

But despite such concerns--hey, new toy. Looks shiny and fun right now. Could be a pain in the ass to actually get it working.

Anyway, I'm planning to run OpenSolaris under VMWare. VMWare, by the way, looks really fucking sweet. I've only used it a little, but... cool. Papers about it, which I read back at UT... cool. It's proprietary, but the VMWare Server Console is beer-free, and has at least a bit to play around with.

So the holdup at this point is the Sun Developer Network, or as the military refers to it, the SDN. (This is a joke; as a fun parlor game, if the military were to use SDN, what would it stand for?) Any similarities in name or intent between the SDN and the MSDN are purely coincidental.

I'm familiar with the general concept of Developer Networks, but have long held them to be tools of the Evil Proprietary Establishment, and have not learned the details, or joined one (which I vaguely considered to be too expensive to contemplate, anyway). But Sun's is free. I think it has recently morphed from something else, upgraded its website, acquired new superpowers, etc. It's pointed to by OpenSolaris.org. But I can't download any fucking software from it.

Rediculous. (And yes, I do know how to spell.)

It's funny how so much of everything comes down to doing things right.

mass on-demand printing 2

There are other problems with small-circulation literary journals, aside from the fixed costs of printing them at high quality: each journal may have trouble coming up with enough quality submissions to publish regularly, and there's not much of a reviewing mechanism to suggest which journals I'm likely to find interesting and should bother reading. Furthermore, there may be some natural subculture of people interested in a particular magazine, but these people probably don't live in the same area (although I suppose if they do live in the same area, they might very well go to the same coffeeshop).

Now, what I'm imagining is that various people publish things on the Web, much like they do today in blogs and on personal webpages, and perhaps in zines and whatever. They make this available under a license that allows it to be included by other people in their magazines. These may be online magazines, but they may also be on-demand print magazines produced as PSFs. The PSF style would also allow a way for people to charge for the use of their pieces on the basis of per-copy royalties. (I.e., anyone may put this in their magazine's PSFs, but every time a copy is printed I should get 5 cents.)

Now, people can create their magazines using these freely available submissions, perhaps adding pieces exclusive to their mag. They add value not only by collecting exclusive submissions but by doing design, layout, copyediting, etc. They also add value by selecting good pieces; in some sense I may follow a particular mag because I share the editor's tastes. Publishers may also sell ads in their mags, which I'll get to below.

If a publisher wants, they can print copies and distribute them locally or mail them to subscribers, but they don't have to. They make the mag available online as PSFs. It can be printed on demand by anyone who wants it.

Now, this clearly addresses the content-scarcity problem, since writers (and photographers, cartoonists, etc.) everywhere are making their material available for use in different mags. It also addresses the problem of a geographically diffuse group of interested readers, since people can find the mag over the Web and have it printed at their local Kinko's (err, their locally-owned mom-and-pop print shop). The reviewing problem, which I will more CS-ically refer to as the filtering problem, is a more difficult one. There are mechanisms developing in cyberspace (and particularly in the blogosphere and web-commerce-o-sphere) to address this problem, such as collaborative filtering (think Amazon recommendations), PageRank and other forms of link analysis, social networks, blogs linking to other blogs, and RSS feeds. All of these techniques can be applied to creators (writers, photographers, artists), to individual pieces, to individual issues of a mag, and to mags themselves.

Two criticisms come to mind that I imagine people will make of this whole "mod p" idea. First, is there really much value to aggregation? Can't I just use the filtering mechanisms described above to direct me to individual authors or pieces, and view them on the author's blogs or websites? Does packaging things as a magazine offer any significant advantage over reading blogs, watching RSS feeds, and following links to pages that look interesting? Second, why bother printing? Even if people want an attractively designed and packaged magazine, why go to the hassle and expense of printing it, rather than reading it online?

In my next post I will take a stab at refuting these criticisms. I will perhaps also get around to discussing opportunities for selling advertising in a (mod p) system.

mass on-demand printing

I was looking just now at a small, low-budget literary magazine, umpteen of which are published and
distributed in colleges, high schools, and coffeeshops across the country. It seems to me that actually publishing one of these, at its most efficient, would involve editing the submissions into a computer file format that a commercial printer could feed more or less directly into its printing systems (taking into account instructions about how to print and bind it, some of which could also be formally encoded). Let's take this system to a logical extreme in which there are machines that accept such files as input and produce complete, printed and bound publications as output. We will call these files "print specfication files", or PSFs for short, and any language in which they are written (a low-level language like PostScript or PDF) a "print specification format" (also PSF).

Note that a PSF represents a clean interface between the creative and mechanical aspects of publishing. It is an interface that reflects the actual division of labor in the production of a small magazine or a corporate newsletter between the editorial staff and writers of the publication, whose job is to produce PSFs, and a (usually separate) commercial printer, whose job is to take the PSFs and produce mass quantities of the publications they describe.

Now, the history of printing economics is all about capital costs and economies of scale. Historically, printing presses have been bulky, expensive, and difficult to operate. Futhermore, printing a document meant manually setting movable metal type into plates; printing then involved dipping these plates in ink and stamping them down onto paper (hence the term printing press). Therefore a huge initial effort was required to set up a document for printing (even after that document had been written and designed to the extent now necessary to produce a PSF), after which an unlimited number of copies could be made at a low marginal cost.

In other words, if I wanted a printer to produce 10,000 copies of a magazine, they would face a large fixed cost of doing so that would not be much higher than that of printing 100 copies, or much lower than printing 100,000 copies. Therefore I will see enormous economies of scale. The revenue I need to generate per issue (via advertising and/or sales) for my magazine to cover printing costs will decrease dramatically as the number of copies I print (and sell) increases.

The practical consequence of this is that in practice it is only feasible to print a financially self-supporting mag pazine at a given level of quality once I reach some critical circulation level, because below this level the price I have to charge per copy to recoup my costs is higher than what people are willing to pay for a magazine. Another consequence is that I need a large amount of start-up capital to print the initial issues of my magazine and distribute them to build circulation. (And such capital is necessary even if creative and editorial staff work for free.)

Desktop publishing was a revolution because the fixed costs of printing things like fliers, newsletters, and smallish magazines (still low quality compared to glossy mass-market magazines or professionally printed books, but acceptable for many purposes) became very low. If I have a laser printer, the cost of printing 100 copies each of 100 different documents will be the same (or only very slightly higher) than the cost of printing 10,000 copies of a single document. This is much, much different from a conventional printing press.

Unfortunately, I don't know the details of modern printing technologies that are higher-end than the kind of laser and ink-jet printers that people have at home or around the office. It seems to me, though, that this new economics of desktop publishing could probably encompass printing and binding a decent-quality glossy magazine, something surpassing the quality of most small-time literary magazines you see today. In other words, the per-issue cost of printing my coffeshop's in-house literary mag should be close to the per-issue cost of printing a mag with much larger circulation at a comparable level of quality.

What I'm getting it is that I should be able to produce a magazine PSF and post it on my website. Somebody else should then be able to find my work on the web, go into their local printer's shop, and have a copy printed out for them, and the price they pay the printer should be around the cost of a conventional magazine they find at a newsstand. This would be mass on-demand printing.

Art Blogging

I think web journals (blogs) are underexploited as a promotional vehicle. There has been a lot of news lately about bigwigs at huge companies like Intel starting blogs that allow them to talk person-to-person with the average Joe employee, which is probably a good thing on the whole. What I'm talking about, though, is promotion for the little guy.

The other night in Austin I was at a gathering of local artists meeting to critique each other's work and offer suggestions; I was invited there by a painter friend. Most people brought paintings; one guy had poems written on index cards that I had originally seen hanging on a coat hanger above his kitchen table, and another had a short video with footage of the World Trade Center attacks and subsequent events, and Radiohead's "Paranoid Android" as the soundtrack. (He says it was booed out of the Alamo as being pro-Bush, which I think is ridiculous.)

In any case, they were planning to have regular meetings and were talking about starting a website, registering a domain name, etc. I think it would be great for them to host blogs. For one thing, it would offer a "day in the life of the artist" sort of view and would be a great venue for promoting their works. They could have regular blogs sort of like magazine columns, or shorter miniseries-style peeks into different parts of the artistic process, or whatever, as well as just letting anyone have their own journal. It would not be too hard for them to design a distinctive and classy look for their website that would stand out from the average blog templates (and they could find local graphic artists, type designers, web designers, etc. to help them).

I think blogs could build the sort of bonds that people with pretentions to connoiseurship would value very highly, and it could even bolster Austin's overall image as a vibrant art scene, at least in a small way.

Blogophobia

Okay, this blogging software sucks. I just oh-so-foolishly hit the "back" button (which, I swear to God, I was once specifically taught not to do in a computer science course at Stanford), and of course I lost my entire draft. That is a risk of doing things over the Web, I suppose...

I really want to host my own blog on my ink08 server, perhaps using the Elite Journal software, since, after all, it is written by a guy named Scott. I actually managed to install Ruby, Gem, and the RubyGems for the Ruby on Rails framework, thanks to some excellent instructions by Bruce Perens on the Ruby on Rails Wiki. I was even able to install Elite Journal from its gems and overcome a bit of a hitch that required me to install some of the SQLite packages from Debian that were dependencies of a dependency. But of course when I run Elite Journal it doesn't work and I get some cryptic Ruby on Rails error message. Back to work another day.

On the other hand, writing my own blogging software would be an excellent way to learn Python, which I need to do anyway. Perhaps I could even provoke some kind of a rivalry with Elite Journal, since I think there's a bit of a thing between Ruby and Python. Perhaps I'll create a "Python Eats Snails" or "Python in Jail" framework to compete with Ruby on Rails.