Tuesday, September 02, 2003

Free speech for all, one byte at a time

Today I got to thinking about how cheap disk storage is now. Internal IDE disk storage is running at around $1 per GB, and external something like $1.4 per GB and the prices scale pretty linearly all the way up to 500GB. That, I would say, is pretty amazing. When I first started computing in the late 1970s a 90k (or even less) floopy drive was $150, and an 8Mb hard drive was just way off the scale. Of course I also remember early encounters with wide area networking at dialup speeds of 300 bps or less. Thats slow enough that you can still read data coming down the line in real time - about five words a second.

So compared with modern day experiences my available bandwidth is up by a factor of a few thousand (1.5Mps), but mass storage is just way off the scale at about one million times what I had just 25 years ago, and for roughly the same dollar cost. I'll tell you, a factor of a million increase is a lot, an awful lot, or as manager would say: "this is insane". When hard drive manufacturers describe what you could do with that much storage they usually say stuff like "store X hours of DVD quality video" or "Y thousand MP3 songs" or "Z hundred thousand high quality photos". Yes those are useful things for the average consumer, but lets face it, in terms of mass storage your average consumer has just about run out of ways to fill disks that big with anything other than backup copies of data on other disks that big.

So when looking at some of the quite lovely and elegant external firewire drives from LaCie I imagined a stack of just ten of their 320GB "big disks" daisy chained together with a 400Mbps firewire connection. For just under $4000 dollars I can buy 3.2 Terabytes of storage. Now that really is insane. Think about it, 3.2TB is 10k of storage per man, woman and child in the USA, or 1/2K per man, woman and child on the planet. 1/2K is just about enough to get your name, age, D.O.B., phone number, ID number if any, and links to relatives information. 10k gets you all of that, plus a reasonably good compressed photo and perhaps a few kilobytes of your own info.

Which got me thinking...

There are companies like Yahoo out there who offer free web site hosting and they have large tracts of terrabytes of storage connected to huge pipes of bandwidth via humming racks of servers that cost millions of dollars. And there is me. The problem with Yahoo and their ilk is a) they put ads and other stuff on your home pages (boo), b) they censor what you put on your site beyond any restrictions required by the law of the land, and c) they are a commercial entity and spend all their time trying to upsell you with more services. So what if I, or any small independent company with a very modest amount of cash decided to set up a free-speech hosting service. The idea being that everyone needs a free and unfettered place to communicate with the rest of the world as they please. The free-speech hosting service gives each person in the country gets one sub-megabyte storage space to do with as they please, for free, no strings attached, no advertising etc. etc. Just have at it.

By my estimation 50% of Americans don't have access to or will never use, the Internet, at least not for a good number of years until my generation, "Generation-I", has moved into retirement. Even if wildly popular at most 10% of eligible users would ever use such free-speech hosting with the limitations it would have. That means my 3.2TB of storage could actually offer an average of 200k to each user, even without allowing for compression. Now that's quite a lot for pure text - my entire blog output (in its natural and very verbose HTML format) is still under 100k compressed, most newspapers would probably fit in 200k compressed without photos. And there is nothing to stop people sharing their unused space with others to create larger community spaces. Basically 3.2 TB provides a lot of free-speech for all.

For a little while I was intrigued, but then the calculations began to reveal a few stumbling blocks. My server, or just a few more like it could probably handle 1000 requests or more per second (1ms each) which is 3.6M per hour or 86.4M per day. Unfortunately my current upstream bandwidth means the first thousand requests from the first second, even if they were to deliver just 20k of data each, would take over 20 minutes to send up the pipe. It would take two and a half years to transfer the 3.2 TB of data just once, let alone handling multiple access per page per day. So with 14 million users the average user would be lucky to download their page in their life time. Oops, that would be quite a world wide wait indeed. I don't think there is an HTTP error for "server became obsolete while serving page", or "client deceased". Even with a thousand times the bandwidth in my pipe, which would be hugely expensive, we'd only just about be making some headway on the bandwidth problem, and never mind peak loads.

So while I'm still thinking "this is insane" that I can, for a very modest sum of money, supply a decent amount of storage to everyone in this country, or even create a "I am here" entry for the entire population of the world, its unlikely that I'll be moving into the free speech hosting market any time soon. One possible solution would be to ask anyone using the service, or the government on their behalf, to fork over a $.25 each one time payment. That's $72 million which should be enough to buy an awful lot of storage, servers and bandwidth and endow a foundation to keep it ticking over forever (using wind and solar power of course). Another solution is just to wait a few more years. If twenty five years can deliver a million-fold increase in storage and thousand fold increase in bandwidth I'm guess I'll just have to wait five to ten years longer before my crazy scheme is feasible. Of course by then everyone will be doing it, or by that time the first-ammedment will have been revoked anyway, for copyright and national security reasons...

No comments: