Doxpara: ZapMail Redux: A Response

"Customer-Owned Networks and Zapmail": http://shirky.com/writings/zapmail.html

It's funny. For quite some time, people have been saying 1984 has come upon us.

It took Clay Shirky -- whose reputation is astonishingly well deserved -- to prove people were right. 1984 is upon us -- for in the year 1984, ZapMail, by FedEx, was brought unto the world. The basic idea was that you'd FedEx your document, only they wouldn't actually send it -- they'd buy a fax machine, and fax it for you. Delivery time would go from 24 hours to 2. Profit!

FedEx lost hundreds of millions of dollars before realizing businesses were just buying their own fax machines and abandoning the per-use charge(not to mention the rather tremendous inefficiencies of having somebody physically transport a document so that it may eventually be electronically transported!)

It is 2003. Instead of fax, we have WiFi. The stakes are larger.

But little has changed.

I hope, perhaps expect -- sadly, I lack the right to demand -- that Clay Shirky's recent article "Customer-Owned Networks and ZapMail" be distributed at business schools across the world, if only for its sheer grasp of historical perspective. Clay gets it.

I've some thoughts on his article (this was originally going to be a simple post on BoingBoing...it grew.):

1) Vonage is indeed awesome -- but the phone companies aren't utterly screwed. Clay claims the phone companies can't or won't compete with Vonage at their own game (fixed price, reasonably affordable flat rate service). Why, they'd go bankrupt!

Hmmm, 'bout that.

MCI, the company that took down Ma Bell, the company now owned by the already quite bankrupt (and thus debt relieved) Worldcom...they've been doing something very interesting. It's called The Neighborhood. It costs between $50 and $70 a month, depending on what state you're in, but that's it -- no local charges, no long distance charges (within the US), all the perks. And it works on your existing phone line.

It's basically Vonage without the IP phone -- or the new phone number. Given the cost of broadband network access, it's either the same cost or even cheaper than the Vonage offering.

Even with pretty low-key marketing, they've managed to rack up over a million subscribers. As with most human consumption, I'm sure usage has increased as the need to track minutes disappeared. As with most all-you-can-eat plans, more than a few people are still paying more per month than than they'd otherwise be paying per minutes.

It's surprisingly likely they don't care.

2) The phone companies made a *fortune* off of faxes. Business got global before Internet Email became a reality, and while faxes beat the crap out of FedEx (thus the attempt to co-opt the product -- again, brilliant explanation), they're really horrendously slow. And let me tell you -- they just chew up minutes. What were long distance rates in the mid 80's? Quarter a minute? Phone companies got quite rich off of faxes.

The Internet doesn't offer similar economics. Large scale bandwidth may not be too cheap to meter -- but individual emails and interactive web browsers are. VoIP, particularly future versions of it, are very likely to be as well.

3) Clay writes: "There may be a small business in wiring "third places" -- coffee shops, hotels, and meeting rooms -- but that will be a marginal business at best." He's right, but I don't think he realized how the rest of his essay explains why. 802.11b access points cost around $100. DSL access, even for businesses, can be acquired for about the same per month. Over the first year, that amortizes to $108 dollars a month...or about $3.60 per day.

How many additional cups of coffee need to be sold to make back $3.60? Lets not even talk about hotel rooms.

The economics aren't there to be the third party...but they're quite nice if you're the first.

Clay got me thinking. If I may be so bold, I'd like to point out another odd similarity. The techno-economic state of Internet companies bears an interesting resemblance to whats happened in recent years to hard drives. As DWDM radically increased the data transmission capacity of fiber, GMR -- Gigantic Magnetoristance, discovered at IBM -- radically increased the data storage capacity of sealed magnetic platters. The data on fiber lines travels long distances almost instantaneously; its bandwidth is spatial. The data on platters travels effectively nowhere, but it stores that data for an eternity(relative to light; relative to humans, a couple years, progressively less). Its bandwidth is temporal.

As fiber has overtaken copper, as TCP/IP has overtaken innumerable proprietary protocols -- GMR overtook everything else, not just throughout hard drive technologies, but throughout the storage world itself. As of now, the only technology that can scale to back up one hard drive is another; a hundred blank CD's struggle to store the contents of a single set of magnetic platters.

In the face of such competition, what else can be done? iVDR -- Information Versatile Disk for Removable Storage -- is simply going to be a hard drive in a removable case. Will it work? Depends if they manage to keep DRM out of it. Without DRM, the potential is enormous. Extremely high capacity, easy to use removable storage will likely become the dominant form factor for all but the most lightweight applications. (Replace "storage" with "network bandwidth", and "form factor" with "access method", and you'll see the analogy survives just fine.)

It'll replace everything but the most speed critical (and now very expensive) server hard drives, and even then only when power and capacity issues prevent the unprecedentedly RAIDable storage capacity from being pressed into service.

With DRM, it'll just be another curious failure, like MiniDisc (which by all rights should have utterly wiped out CD's and Cassettes), NetMD (which does one thing, and does it barely, and would rather turn into a steaming slag of molten metal than do anything else), and Press...I mean DataPlay.

Serve few masters, make few dollars. Serve the masters of the market -- consumers -- and make many more.

That's not to say that fixed storage is doomed. Your system probably spends 90% of the time reading (not writing) the same 512MB-1GB of hard drive space. RAM -- particularly PC-133 -- is cheap. Eventually, Maxtor or Western Digital (IBM's up to something else) will wake up and start selling battery backed, RAM based boot drives. There's already one out on the market, but it's immature and expensive. RAM is several orders of magnitude faster and quicker than spinning platters; provided most transfers fit within a newly massive buffer, the temporal cost of setting up data transfers should approach zero.

How is this familiar? The following runs the risk of violently overextending an analogy, but bear with me -- this is turning out vastly more rational than I expected. WiFi may be slower -- less spatial bandwidth, as it were -- but it's more than fast enough. Meanwhile, the ease in which new transactions may be initiated is unparalleled. When it works right, no babysitting, no hard work -- you've got net the moment you pull your laptop out of its slumber (provided there's no security in place, at which point even if you're a legitimate user, you're likely to give up. Hello again, DRM). Get in, get out, everything works nearly instantly.

Primary storage through batteried solid state RAM -- possibly backed by platters, possibly by compact flash, maybe by nothing at all (Palm-style) -- has interesting potential. I imagine it takes less power to pulse the life into DRAM than it does to spin a hard drive and move a read head, so I'd expect to eventually see Laptops adopt this strategy (*cough*). They've already got batteries, they're already pulsing RAM, and they already own the entire motherboard platform. All they'd need to do is support more memory, allow their memory controllers to expose a configurable amount of space as a bootable IDE drive, and see what users do with it. Provided a modicum of hardware support for hibernation, backing up onto the platters is already supported.

Technically, this could all be done in Linux, *today*, entirely in software, minus the reliability of effectively a hardware journal (something I'm sure more than a few caching controllers implemented back in the day). Can hdparm (the Linux tool for hard drive control) completely cut power to a hard drive? Hacking minds with insufficient battery life want to know.

Yours Truly,

P.S. Thank you, Clay. I'll inform the Gnomes that Step #2 has finally been documented.