What do trends in the climate change movement mean for a decentralized internet?
Al Gore’s talk from TED 2016 was just published, and it’s worth watching:
I won’t steal Mr Gore’s pitch, as it’s important that people hear the whole thing, so I’ll ask you to watch it, then come back.
As you now know, two very hopeful economic trends are now under way: the rapid drop in cost of renewable energy (in particular solar power), getting close to a tipping point where it may be cheaper than using gaia-toxic energy sources, and the equally important drop in prices for energy storage devices (batteries).
Listening to this talk made me excited and hopeful for the planet, but it also made me wonder about the potential shifts that these trends will have on other critical systems, such as the Internet.
Solar energy is decentralized energy
In particular, solar energy and batteries have a remarkable property: they need to be distributed near where people are. There is a structural disadvantage to centralizing this kind of energy: because of power losses over large distances, given equal cost-per-watt of energy capture, it’s more efficient to do the power collection and storage close to where people need to consume it.
I don’t think we yet have found the comparable approach when it comes to the mainstream consumer internet (where the economic advantage seems to be strongly towards centralization), but I find the energy transition that Gore articulates provocative, and one that I want to learn more about. There may be lessons there for those of us who worry about an over-centralization of economic and political power in the digital realm.
What internet should we build on a decentralized energy infrastructure?
If Gore is correct that solar cells and batteries are key parts of a transition to a sustainable energy future, it’s worth noting that these same technologies can not only be used to light and heat homes, but also to power computing devices, from phones to computers to routers to transmission equipment. So, computing will be possible at the edges. How can we turn this into connected computing?
If one looks at the internet ‘at the edges’ from a perspective of a western developer, it feels fairly developer-hostile. The cloud infrastructure that we’ve gotten used to is very far from the user’s devices. Latencies, bandwidth, even reliability of connectivity between where we think computing happens (AWS and the like) and the potential users make the fit between our approach and that audience feel very awkward.
It may be provocative however to look at this problem the other way around: start with the notion of an internet node as one that is at the edge, and think about what technological advances would be needed to make it a valuable resource to that user, what connections it needs, and what kinds of communication systems can power it.
For that node to be useful, it needs to have access to relevant information, connect with relevant people. Often, those will be relatively local. And the assumption of latency at the edges is pretty bad, so real user benefit can result even by “outrageously slow” performance using, e.g., California benchmarks.
This makes me think about propagation networks and p2p systems. It makes me want to ask whether there are addressing and data access models that take into account the effective topology of a network. It makes me wonder who is pushing the envelope of these sparse connection environments. And it makes me think that we need someone like Gore to tell that story and inspire a next generation of technologists who will see the frontier not as a barren landscape but a place full of potential.