>5/ A decentralized internet will emerge, led initially by decentralized infrastructure services like storage, bandwidth, compute, etc. The emergence of decentralized consumer applications will be slow to take hold and a killer decentralized consumer app will not emerge until the latter part of the decade.
The pendulum of history suggests this will occur (at some point), and I hope it happens sooner than later in many respects, but it is also seems like one in which we won't know the triggers/causes/sparks until after the fact, partially because it seems it will take complex combinations of causes?
Anyone seeing possible sparks which perhaps the rest of us aren't yet identifying?
Maybe there will be a need for massive computing in remote areas: Antartica, or space. They need a lot of local storage and compute. And they have low bandwidth.
It's kind of like GPUs are in cars right now. You can't drive a Tesla with dumb sensors over the Internet -- you need smart local compute.
So I guess IoT and doing heavy local computation is a technical reason you would need decentralization. I can see that happening for many use cases. I'm not sure if it will happen for the consumer web because centralization is more efficient and the current network effects are so ingrained. Similar to how Windows is still dominant on the desktop, but iOS/Android are perhaps more important platforms.
---
I think major changes in behavior are driven by new hardware -- phones in the 00's, PC's in the 80's, Internet in the 90's, etc.
People have been trying to push VR, but to me VIDEO is the real VR -- more stuff happens there and more people use it. I was chatting with a friend yesterday and observed that YouTube is basically what "SecondLife" was supposed to be. People are exchanging all kinds of valuable information and entertainment on YouTube.
So if you need to process a lot of video locally for some reason, that could be a killer app for decentralization. Just like a self-driving car, although I'm bearish on self-driving impacting the average consumer in the next 10 years. I think it will continue to be cheaper to operate rideshares with human drivers in most parts of the world and most terrains/climates.
Yeah it's probably because I've been watching a lot of YouTube lately, but it feels like there's just a lot more real interaction going on there than on other platforms. It sounds like Twitch is the same way.
One example: I learned how to clean my toilet from this video.
The comments are hilarious... Tons of people having the same "AHA" moment. (Basically you paper mache your toilet with vinegar and wait a couple hours. Old mineral stains come off like butter!)
Compare a google search for "clean toilet" and it feels like a bunch of SEO-infested crap.
YouTube is more like the "old web" where you can get a real opinion on something.
-----
I have friends who cook and that's a whole other subculture of YouTube. I've been watching a good MMA show. And there are programming streams, and pretty much every programming conference has an archive, which is a rich archive of free information (e.g. PyCon, CppCon, etc.)
I don't know what's going on in Second Life now but to me it feels like it's probably not "real life". I guess people want "life" and not "second life", and video is becoming an increasingly large aspect of the former.
Rebutting my own comment: even if you need heavy computation and storage locally, that still doesn't motivate a decentralized network.
I would think of it as control plane vs. data plane. The data plane can be massively distributed in space, but the control plane can still be centralized.
And of course that's how Tesla works, and how software-defined networking works.
The "powers that be" just need to control software updates and the network's control plane (routing). They can remotely manage distributed resources.
So yeah unfortunately I'm not seeing a big motivator for decentralized networks (which can be very, very slow). You would have to have some need for a lot of local video processing but also a whole way to distribute code and software updates.
And right now that's more centralized than it's ever been. I'm not a fan of the "silent, frequent, and huge updates and pop up new TOS" model but that's the status quo.
The spark is already here. I work in a rapidly emerging domain where the trends clearly indicate that traditional concepts of centralized infrastructure cannot serve the required workloads: operational sensor/geospatial data models. Basically, machinery measuring and reasoning about the physical world at scale, often in real-time. Several aspects of these data models (technical, economic, regulatory) strongly indicate a globally federated implementation that allows for fast, decentralized, ad hoc cooperation of storage-dense compute elements at the edge. The aggregate data velocity is so high that the physics of data model centralization is untenable, so there is a certain near-term inevitability about it even though you can make a centralized solution work today.
There is active research into the theoretical and practical design of systems and protocols that will make this plausible. It has no precedent in literature and it is a very non-trivial problem but the sense is that a practical workable design is achievable in the not too distant future.
It is worth noting that effectively managing climate change requires implementing the same kind of data model with similar theoretical constraints. Building data models of physical reality at scale breaks just about every part of classic data infrastructure architecture.
There will be another factorial increase in, for a lack of a better term, email attachment sizes.
It's still hard to share files that are 500MB in size, and I don't see why. I think it has to do with media companies like Google not wanting individuals to share files, unless it is through them. But the damn will break soon, much like Megaupload changed the scene in 2005.
ipfs, dat, zeronet I think are good examples of the sparks you are looking for.
These are outside of the blockchain world of compute/storage as a service attempts that got started suring the ico goldrush and seem to be doing quite well for themselves.
i might categorize protocols (or even combining ipfs and dat as the basis of interesting solutions) as fuel. not sure they are the spark that lights the fire.
maybe that sounds like semantics, so to propose a rough taxonomy of different types of actors:
a. nation state level superpowers
b. nation state level challengers
c. large business / incumbents / leaders
d. small business / startups / challengers
e. individuals / consumers / social groups
f. possibly horizontal groups across combinations of the above
it would seem at least one of those groups would need to believe they can reap move-the-needle level benefits from decentralized internet in order to spark progress?
I agree. They are just modernized torrents with higher usability. The real problem (consensus) still only has one solution: blockchain with proof of work.
I think we should be asking: What problem does it solve? why is that problem important? and why would someone use that new solution rather than what they're already using.
The pendulum of history suggests this will occur (at some point), and I hope it happens sooner than later in many respects, but it is also seems like one in which we won't know the triggers/causes/sparks until after the fact, partially because it seems it will take complex combinations of causes?
Anyone seeing possible sparks which perhaps the rest of us aren't yet identifying?