- Blog Home
- Jacob Kring
- On Distributed Systems
On Distributed Systems
The Question: Distributed systems (like the internet) tend to be more stable and efficient than centralized systems (like a filing cabinet). So, why are big department stores putting all the mom and pop shops out of business?
The Hypothesis: The benefits implied by "economies of scale" could be outweighed by an efficient and distributed information system.
Bob needs a tennis racket, a pound of flour and a box of tissues. He doesn't actually need a full box of tissues, or a full pound of flour for that matter, but his plan for the day is to blow his nose, bake some cookies, and bring them to the tennis courts for his tennis date. He's never played tennis before, and the sad irony is that while he is very excited for his first tennis experience, he is going to hate it. Like golf, tennis requires a patient demeanor that Bob simply does not possess. After today, he will never play tennis again.
Bob goes to Walmart because he knows he will find tennis rackets, flour and tissues there. Walmart is a 20 mile drive from his apartment. His car gets 20 miles to the gallon.
When Bob gets to Walmart, he buys a new tennis racket, a pound of flour, and a box of tissues.
Bob is the unfortunate, unknowing victim of an inefficient and centralized system.
Sally, his neighbor from across the hall, just got over her cold and has an extra box of tissues that she never opened. Tom, the guy in the apartment above Bob's, just started this new vegan diet. He's got a log of Pillsbury Cookie Dough in the refrigerator that's slowly eroding his morale. He can't, in good conscience, throw the food away, but he also can't stand to look at it every morning as he drinks his spinach and almond milk smoothie. Just this morning Jen, from across town, schlepped all her old tennis gear to Second Handle, the used sporting goods store down the block from Bob's apartment building.
Sally, Tom and Second Handle are all nodes in a distributed system. The problem is they are not yet linked. And a distributed system with nodes that don't communicate isn't really a system at all. Even if all those nodes were connected, Bob would still end up at Walmart if he weren't somehow connected to them as well. The good news is: all of this is merely an information problem, and over the past 30 years we've done a pretty good job of solving those. Which brings me to one of my biggest concerns: with the growth of the tech sector, aren't we simultaneously growing our economy's capacity for wealth disparity? If, with every open-source release, a smaller team can build a bigger company with greater scale, aren't we on the fast track toward a world in which the 1% is actually the .0000000001%?
Thankfully, the dream of a distributed system assuages this concern, as well. This is because, by decentralizing commerce, the system is similarly decentralizing wealth. Instead of Bob's money going to Sam Walton, in a distributed system it would go to Sally or Second Handle. Sure, there will still be big companies. But those companies will merely facilitate the flow of information, and they'll profit from trivial transaction fees. With the rise of open-source, that will be a race to the bottom, in fact some argue it's already approaching zero. Who knows, maybe with a fully functional distributed system we could even shed the shackles of debt and return to a virtual, credit-based economy.