Back in the day, before the WWW, we had the first networks (barely). What did we want to build? The World Computer.
When the WWW booted up, what did we want to build on top of HTTP? The Internet Computer.
And we actually have built things kind of like A World Computer. SETI@home. Condor. Clouds.
So it’s no surprise to an old gray hair like me that when people see a blockchain, they want to build: The World Computer.
And, indeed, “smart contracts” a la Ethereum are sort of, kind of, a global computing facility.
But like all previous World Computer concepts, blockchains haven’t delivered yet.
Personally, I’m not holding my breath. Operating a large-scale computing system is really complicated and hard. (I helped do this for many years, so I know what I’m talking about. )
And, honestly, the Internet is a terrible platform to try to build on. Heterogeneous. Buggy. Unreliable. Out of control. Weird economics. Unknown and unknowable legalities.
A blockchain is layered atop the Internet, adding complicated decentralized and asynchronous protocols. All of the above plus more latency, opacity, and unmodifiable write-once processes.
Yeah, that sounds promising.
So I read this summer of yet another shot at this concept, Dfinity [2]. Actually, the first I heard of this project was a troubling headline, “The Dramatic Crash of a Buzzy Cryptocurrency Raises Eyebrows”, referring to the sudden implosion of their token just weeks after launch [3].
“Even in the famously volatile crypto market, [Dfinity] stands out”
(from [3])
It’s over before I even heard of it, and certainly before it had a chance to prove itself.
Obviously, there has been plenty of chit chat about this billion dollar swan dive. Many people note that there was suspicious trading that favored insiders, combined with restrictions on small investments (i.e., ‘outsiders’), and some Reddit mischief to help things along [1].
But what did they think they were doing, and why was this supposed to be different this time?
Looking at the technology, Dfinity gets full marks for complexity. One of the big ideas is to try to reproduce the concept of virtual machines in the form of essentially virtual blockchains. This is kind of interesting, letting different organizations and projects have their own “sub net”, a blockchain with parameters that work for them. Eliminate the “one size fits all” problem, and potentially improve performance by keeping the relevant network to reasonable scale and locality.
Now, all these little domains are supposed to be part of a larger tapestry, potentially interacting in sensible ways. I don’t fully grok this part, but they seem to be dreaming of possibly composing specialized chunks into larger applications.
How would all this happen? Dfinity works like an operating system, specifically, like a cloud management system. It manages these virtual sub groups, assigns work, and apparently manages the physical resources that run the virtual blockchain nodes.
This is pretty much what a big cloud does, except Dfinity doesn’t “own” the hardware, it negotiates with independent nodes via Nakamoto-style protocols. Computing resources run Defnity, and are assigned to some sub net. Dfinity oversees the behavior of nodes, detecting dropouts, reassigning work, and otherwise running the show.
This is standard stuff for distributed operating systems, except Dfinity is doing it Nakamoto-style. Decentralized. (“Damn your eyes! I said decentralize!”)
It’s all crazy complicated, to the point that I seriously doubt that anyone has any idea how it would really work.
For example, a classic Nakamotoan blockchain has one synchronization / consensus protocol, for better or worse. Dfinity lets you use one of three (I think).
So, if one unicycle is hard to ride, let’s ride three unicycles. That will be much easier!
For me, the “technical” documentation was revealing because it has extremely detailed description of the funding rounds, and moderately detailed explanation of the economic model (i.e., the ‘coins’), and extremely shallow description of the actual technology. And, as usual, there is no refereed publication or independent verification of the software.
But there definitely are interesting unverified claims such as:
“At genesis, the IC will have a block rate of one block per second (bps), then move up to ~1,000bps by this year-end. According to Williams, there is theoretically no upper limit to blocks per second.”
(from [2])
Hmm. Infinite speed? On top of Etherum running on the regular Internet?
Right.
(By the way, the “no upper bound” assertion is supported by a citation—to a tweet.)
I also love statements like, “The ecosystem will find a natural balance of how many rewards are required to achieve an optimum amount of participation.” [2]. (Insert animation of a hand wildly waving here.)
Wanna bet? (Judging by the 95% loss in value [3], apparently a lot of people wanted to bet “no”.)
All in all, I would predict that, should it ever fly at all, Dfinity will have trouble getting enough users to be stable and secure. It will probably experience that particular hell of being too expensive for users, but at the same time losing money for operators. Everybody loses. (Except the insiders who already cashed out, of course.)
From the complexity and opacity of the technology, I’d say there is a reasonable probability of a serious bug or security flaw. For that matter, there is a reasonable possibility that it would crash in response to an unexpected event, e.g., a big Earthquake somewhere, or a regulatory change.
So watch out.
However, Dfinity surely must be a candidate for CryptTulip of the Year, in only because it promises infinite performance.
- Arkham Intelligence, REPORT ON THE INTERNET COMPUTER TOKEN. Arkham Intelligence, 2021. https://arkhamintelligence.com/icp/report.pdf
- Mira Christanto and Wilson Withiam, An Introduction to Dfinity and the Internet Computer, in Messari, May 10, 2021. https://messari.io/article/an-introduction-to-dfinity-and-the-internet-computer
- Ephrat Livni and Andrew Ross Sorkin, The Dramatic Crash of a Buzzy Cryptocurrency Raises Eyebrows, in New York Times. 2021: New York. https://www.nytimes.com/2021/06/28/business/dealbook/icp-cryptocurrency-crash.html
Cryptocurrency Thursday
Robert McGrath's Blog 