Monday, March 26, 2018

Replying to why decentralization matters

Last month Chris Dixon published "Why decentralization matters".

I felt it was one-sided and lacked solid arguments. I picked up some populist vibes as well.
"There go my people. I must find out where they are going, so I can lead them." --Alexandre Auguste Ledru-Rollin

At that time, I had jotted down my criticisms to some of the paragraphs in that article in my draft posts folder. Then I saw Todd Hoff's (from High Scalability fame) brief write up about the article. It captures nicely my reactions to the article. So I will start with Todd's response and cover only the remaining parts of my criticisms in the rest of this post.
"Nerds always dream of decentralization, I certainly do, but every real world force aligns on the side of centralization. We still have NAT with IPv6! Ironically, the reason given why decentralization will win is exactly why it won't: "Decentralized networks can win the third era of the internet for the same reason they won the first era: by winning the hearts and minds of entrepreneurs and developers." Decentralization is harder for developers, it's harder to build great software on top of, it's harder to monetize, it's harder to monitor to control, it's harder to onboard users, everything becomes harder. The advantages are technical, ideological, and mostly irrelevant to users. Wikipedia is crowd-sourced, it's in no way decentralized, it's locked down with process like Fort Knox. Decentralization is great for dumb pipes, that is the original internet, but not much else. Cryptonetworkwashing can't change architectures that have long proven successful in the market."

The myth of the decentralized Internet

Chris says: "During the first era of the internet, internet services were built on open protocols that were controlled by the internet community."

Open protocols do not imply a decentralized system.

Also the myth of decentralized Internet (or early Internet) needs to die already. "The Internet is better viewed as being distributed, both in terms of technologies and governance arrangements. The shift in perspective, from decentralised to distributed, is essential to understand the past and present internet, and to imagine possible future internets which preserve and support the public good."

The excellent book "Where Wizards Stay Up Late: The Origins Of The Internet" tells the story of how ARPA and the involved universities built the origins of Internet and captures really well the thinking/perspectives at that time.

Render unto GAFA

Chris says:
> During the second era of the internet for-profit tech companies, Google, Apple, Facebook, and Amazon (GAFA), built software and services that rapidly outpaced the capabilities of open protocols.
Yes, this is because [logically] centralized coordination is effective and efficient.

> For 3rd parties, this transition from cooperation to competition feels like a bait-and-switch. Over time, the best entrepreneurs, developers, and investors have become wary of building on top of centralized platforms. We now have decades of evidence that doing so will end in disappointment. In addition, users give up privacy, control of their data, and become vulnerable to security breaches. These problems with centralized platforms will likely become even more pronounced in the future.
The first part is a fair point (below I talk a bit about what could be a better model). But I am not sure I agree with the second part. Centralized services provide easy to use platforms for third parties to reach to millions of people that use the service. IoS appstore, Android appstore, AWS, Facebook, and Google ad platforms are examples. These platforms are sharing a great resource/platform to bootstrap and scale to millions to the 3rd parties, and therefore can afford to leverage on that and be whimsical and demanding. This comes with freemarket cut-throat capitalism. I am very disappointed (but not surprised) with how this is abused and I hope we can converge to a better model that prevents the abuses. Hopefully some of the ideas from the decentralized systems can be adopted to help converge to that model.

That said, I don't think decentralized systems would be immune to abuses and tyranny, only the type of power wielders will change.

> Decentralized systems start out half-baked but, under the right conditions, grow exponentially as they attract new contributors.
It is still unclear to me how a decentralized network can bootstrap easily. Decentralized does not automatically resolve the bootstrapping and traction problem, right? Why would I join a decentralized system in the beginning? I don't know how many Byzantine nodes are out there? At the beginning of the network, it is easy to have a Byzantine gang trap. I think the cryptocurrencies play to the greed and FOMO of the people while there is hype in this field. Initially it is cheap to join, and there is a chance to bank on by being the early adapter and make money off the late comers. But that works only if this particular thing/platform becomes popular. That is not a sustainable model. For the dozens of cryptocurrencies that become popular, we have thousands of cryptocurrencies that tanked which are missed. I think we have a case of survivorship bias in evaluating blockchain bootstrapping process.

Appeal of the emotional appeal

Chris says: "Decentralized networks can win the third era of the internet for the same reason they won the first era: by winning the hearts and minds of entrepreneurs and developers."

If you  make an emotional appeal to substitute for a logical argument, you are not doing well. On a side note, recently I had watched a speaker at an important venue that made the following appeal for decentralization: "The cloud took away our personal computers and it is trying to move us back to the mainframe age. With blockchain and peer-to-peer computing we will take back the control to our computers." Oh, please, cry me a river.

As for the response to Chris, Todd's assessment is right about the developers' perspective: "Decentralization is harder for developers, it's harder to build great software on top of, it's harder to monetize, it's harder to monitor to control, it's harder to onboard users, everything becomes harder. The advantages are technical, ideological, and mostly irrelevant to users."

Since the [logically] centralized cloud model has been around for a long time, it has done a lot of improvements in its developer support. Thanks to the logical centralization, the model is also easier for the developers to build on and provide high-availability and virtually unlimited high-scalability as well. But for the decentralized systems, the current support is bad, and there is no indication it can improve fast quickly. 

I think what attracts developers to fully-decentralized platforms currently is not that their minds and hearts are won, but it is the novelty, curiosity, and excitement factors. There is an opportunity to make a winning in this domain if you are one of the first movers. But due to the volatility of the domain, lack of a killer application as of yet, and uncertainty about what could happen when regulations hit, or the bubble burst, it is still a risky move. 

> Cryptonetworks combine the best features of the first two internet eras: community-governed, decentralized networks with capabilities that will eventually exceed those of the most advanced centralized services.
This is another big unsubstantiated claim. Chris says this as a matter of fact and as inevitable without providing any justification for it. I am not convinced that a fully decentralized architecture can be made to exceed the capabilities of [logically] centralized cloud services. Instead I think the centralized solutions can quickly copy the best parts of the technology and outpace decentralized solutions. Paypal can just implement the smartcontracts as the good idea, and still do this as a centralized solution, and avoid all the problems of decentralization. Maybe they can throw in attestation and verifiable claims exchange, which they can more easily implement in their logically-centralized platform. Linked-in may implement attestation and identity services. And identity services will come to social network platforms inevitably/eventually, given the current rate of manipulations and propaganda going on. Dropbox can easily add content-based addressing and leave IPFS in obscurity. Kickstarter can easily implement the token/ICO idea as a better way to support projects with stakeholders.

> Today’s cryptonetworks suffer from limitations that keep them from seriously challenging centralized incumbents. The most severe limitations are around performance and scalability. The next few years will be about fixing these limitations and building networks that form the infrastructure layer of the crypto stack. After that, most of the energy will turn to building applications on top of that infrastructure.
As for the efficiency limitations of decentralized systems, it may not be possible to fix the limitations of decentralized systems. Decentralized is at an inherent disadvantage from a coordination perspective.

What is a better model for preventing abuses with centralized services?

The hand of the free market helps somewhat for keeping centralized services straight: when one big corporation starts playing foul and upsets the users, new companies and startups quickly move in to disrupt the space and fill in the void. But the hand of the free market moves slowly.

This is not exactly my domain of expertise, but here are some things that may help prevent abuses from centralized services.

1. The services can provide some open core protocols/contracts/APIs that won't change. This will give these services an edge for adoption by developers compared to others. Maybe they may even make these enforcable by law with some penalty. Cloud providers offer SLAs, maybe they should offer these contracts to developers as well.

2. Users and developers can be made stakeholders via initial token offerings. Why can't a centralized service not be able to offer tokens? As long as there is a real cost of getting a token (the user and developer makes some contribution to the growth of the platform) and there is eventually going to be real value associated with a token, this incentive economy can also work for centralized services. Badges are a form of token, and it worked OK for a long time for some services for FourSquare for example. Why not take this to the next level with some cryptotokens, and tokens that can be exchanged/sold on and off the platform?

3. The services may be build closer to Unix philosophy of specialized components with better defined inputs/outputs. This can enable third party services to integrate transform. And collaboration and cooperation may bring win-win business to each service involved as well. Why do the services need to be close walled gardens?

4. Finally better regulations from law-makers can help, such as EU's laws about customer privacy and the right to be forgotten.

I realize number 3 is a bit far-fetched. But if we could have it, we could use our contributions and our data to have some leverage over the services. I had thought about this problem for the smartphone crowdsourcing model back in 2013, but did not follow up because it is a very ambitious undertaking.

Other MAD questions

1. What if the decentralized approach wins a decisive victory quickly, what would that look like?
I find this unlikely but let's say this is a black swan event: somehow the decentralized approach click really well with developers and users alike and takes off. And we see more and more of the blockchain based services. What does that world look like? Does that mean there are more stakeholders in hit services. I am all for spreading the wealth. If blockchain wins, is it automatically guaranteed that we have more stakeholders and more accountable services. Or would we still see relatively few number of big money stakeholders dominating the ecosystem? How do you convince a skeptical this would not be the case?


2.  What are some very good applications for PoW blockchains? (I had the same question for IPFS recently.) Chris doesn't give any killer application, but insists that the applications will come. Even if we still miss the killer application, at this point we should be able to name some reasonably good fit applications there, and name applications from the general territory. Here is a checklist to eliminate some answers in advance.
  • It is store of value against the devaluating currency of developing countries. (They have been using gold, USD, Euro for that for many decades thank you.)
  • It is for international money transfer. (Does it beat Paypal, Western Union, banks, etc., significantly in the features that consumers value the most?)
  • But blockchains have this feature and that feature. (What is the killer application? Let's talk applications not features.)
And while we are at it, I repeat Todd's plea. Let's not call Wikipedia a decentralized system ever again. Crowdsourced does not imply decentralized, and certainly not the permissionless fully-hyperbolically-decentralized systems suggested in blockchain PoW approach.

Related links

The comments under Chris's tweet includes a lot of interesting counterpoints and very valid concerns.

The comments under Chris's medium blog post also has several counterpoints presented.

Blockchain applications in Supply Chain Management

Paper review. IPFS: Content addressed, versioned, P2P file system

Paper review. Blockchains from a distributed computing perspective

No comments: