Tuesday, July 3, 2018

Special Report on Decentralizing the Internet (Updated)

The Economist's June 30th issue features a special report from Ludwig Siegele entitled How to fix what has gone wrong with the internet consisting of the following articles:
I really like the way The Economist occasionally allows its writers to address a topic at length. Siegele provides a good overview of what has gone wrong and the competing views of how to fix it. Below the fold, my overall critique, and commentary on some of the articles.

I pointed out in It Isn't About The Technology that the reason the Internet was dominated by the FANG (Facebook, Amazon, Netflix, Google) wasn't that the underlying technology was centralized, it was the economic combination of network effects and economies of scale. Early in the report Siegele appears to agree:
So the first priority is to attract as many users as possible, which today’s online giants did by creating a great “user experience”. ... At some point, Google and the other tech titans also came to benefit from good old economies of scale.
But in the rest of the report there's no evidence that Siegele, like the developers of decentralized systems and applications, understands what I wrote about in 2014's Economies of Scale in Peer-to-Peer Networks, and what the VCs and large companies investing in "blockchains" understand: decentralized systems are also subject to economies of scale, and thus if successful become centralized. I wasn't a prophet. Earlier that year, 5 years after Nakamoto's code release, Bitcoin mining had become centralized:
Researchers from Cornell University say that on multiple occasions, a single mining pool repeatedly contributed more than 51 percent of Bitcoin's total cryptographic hashing output for spans as long as 12 hours. The contributor was GHash, which bills itself as the "#1 Crypto & Bitcoin Mining Pool."
Mining power 25 June 2018
Once this became public miners fled GHash to avoid crashing the value of Bitcoin, but this wasn't really a fix. Four years later two pools apparently owned and operated by Bitmain controlled 41.2% of the Bitcoin hash rate (BTC.com 24.8%, AntPool 16.4%). Add in SlushPool's 10.6% and you have 51.8%. Not very decentralized, and all too easy to mount the kind of 51% "double spend" attack that is becoming routine among smaller "alt-coins".

I know of no effective technological push-back against the centralizing effects of economies of scale in decentralized systems, nor any serious research effort into finding one. ASIC resistance has been a predictable fiasco. Ethereum is trying to fix the inefficiency of Proof of Work with Proof of Stake, which is even more centralizing. Bitcoin is trying to fix the same inefficiency with the Lightning Network, which is centralized. Without a way to prevent centralization, none of the supposed advantages of "blockchains" or decentralized systems will happen in practice, because if the systems become successful enough to make a difference they will no longer be decentralized.

Blockchain developers, like the ones I hung out with at Brewster Kahle's Decentralized Web Summit last year (a much bigger successor is happening next month), are working to create a decentralized Web. You would be naive to believe that the VCs backing these efforts do so in the hope of profiting from the world of small, independent, nodes competing with each other that the developers want to build. No, the VCs expect that the "decentralized Web" will centralize around a few "unicorns" and that they will have funded one. Even if that doesn't happen, they hope that one of the tech giants will buy the not-yet-unicorn. Either way the result is centralization.

But the VCs are not stupid.  They are structuring the ICO deals to eliminate any risk of losing money in the process. Take Filecoin's $257M ICO as an example:
The venture funds who put up the initial $50M, including Union Square Ventures, Andreessen Horowitz and Sequioa, paid less than even the early buyers in the ICO. The VCs paid about $0.80, the earliest buyer paid $1.30. Filecoin is currently trading in the futures market at $7.26, down from a peak at $29.59. The VCs are happy, having found many "greater fools" to whom their investment can, even now, be unloaded at nine times their cost. So are the early buyers in the ICO. The greater fools who bought at the peak have lost more than 70% of their money.
So the funders don't actually care that much whether Filecoin ushers in the nirvana of decentralized storage or not, they've already won.

How to fix what has gone wrong with the internet

Siegele starts by describing why there is so much dissatisfaction with the current Internet:
At the same time the internet has become much more strictly controlled. When access to it was still mainly via desktop or laptop computers, users could stumble across amazing new services and try many things for themselves. These days the main way of getting online is via smartphones and tablets that confine users to carefully circumscribed spaces, or “walled gardens”, which are hardly more exciting than television channels. ... Another control point is cloud computing, which by its nature puts outsiders in charge of applications and their associated data. Meanwhile governments, which long played no part in the internet, have established power over large parts of the network, often using big internet firms as willing enforcers, for instance by getting them to block unwelcome content.
Then Siegele introduces the distinction around which he organizes much of the rest of the report:
The centralisation of the internet and the growing importance of data has given rise to what Frank Pasquale of the University of Maryland, in a recent paper published in American Affairs, calls a “Jeffersonian/Hamiltonian divide” among critics of big tech. One group stands in the tradition of Thomas Jefferson, one of America’s founding fathers, who favoured smaller government and less concentration in business. Its members want to rein in the tech titans through tougher antitrust policies, including break-ups.
The other group are Hamiltonians, who:
argue that without the free services and easy-to-use interfaces offered by companies such as Google and Facebook, far fewer people would be using the internet. Without cloud computing, which lets firms crunch vast quantities of data, AI would be nowhere. Having a few powerful firms in control also helps curb the demons of decentralisation, such as cybercrime and hate speech. This kind of thinking, long used by online giants to make the case against regulation, has gained some traction on the left in the West. But it is mostly thriving in China, where the government wants tech titans to help it in its quest to turn the country into a cyber-superpower.
I would class Siegele as a skeptical Jeffersonian, just not as skeptical as I am. He is too taken with the current hype around "blockchains" (i.e. whatever I'm trying to sell you right now) He writes:
Yet among Jeffersonians a sense of a new beginning is also in the air. The buzz at technology conferences today is reminiscent of 1995, shortly after the birth of the world wide web, when a new piece of software called a browser took the web mainstream, and the internet with it. At today’s events startups are pushing ambitious plans, often based on blockchain technology (immutable distributed ledgers of the sort that underlie Bitcoin and other crypto-currencies), promising to “re-decentralise” the online world.
The distinction between Jeffersonian and Hamiltonian approaches to the problem is useful, as we shall see below.

Blockchain technology may offer a way to re-decentralise the internet

Siegele recognizes that Jeffersonian decentralization isn't a new idea:
To be sure, similar ideas have been tried before—and failed. Decentralised services, then called “peer-to-peer”, briefly flourished in the late 1990s and early 2000s. They fizzled out mainly because no one knew how to build a robust decentralised database. That changed in 2009 with the invention of Bitcoin and the blockchain, the technology that underlies the crypto-currency.
But he's wrong about the cause of the failure. The MIT Media Lab's Defending Internet Freedom through Decentralization report is much more perceptive; it shows that the earlier efforts failed because they were hard to use and didn't offer any compelling advantages over the centralized systems. People like convenience more than privacy – so no, blockchain will not 'decentralise the web' by Matt Asay makes the same point:
It's not that a blockchain-based web isn't possible. After all, the original web was decentralised, too, and came with the privacy guarantees that blockchain-based options today purport to deliver. No, the problem is people.

As user interface designer Brennan Novak details, though the blockchain may solve the crypto crowd's privacy goals, it fails to offer something as secure and easy as a (yes) Facebook or Google login: "The problem exists somewhere between the barrier to entry (user-interface design, technical difficulty to set up, and overall user experience) versus the perceived value of the tool, as seen by Joe Public and Joe Amateur Techie."
And no, in the real world blockchain doesn't solve anyone's privacy goals, as more than 40 dark-web drug dealers just discovered.

Siegele notes that:
Another field of much endeavour is decentralised digital storage. One such effort is Solid, a project led by Sir Tim, which features individual “data pods” where people keep their information (though it does not use blockchain technology). Another is the InterPlanetary File System (IPFS), the brainchild of Juan Benet, a co-founder of Protocol Labs, a startup.
Solid is one of the systems studied in the Media Lab's report:
The approach of Solid towards promoting interoperability and platform-switching is admirable, but it begs the question: why would the incumbent “winners” of our current system, the Facebooks and Twitters of the world, ever opt to switch to this model of interacting with their users? Doing so threatens the business model of these companies, which rely on uniquely collecting and monetizing user data. As such, this open, interoperable model is unlikely to gain traction with already successful large platforms. While a site like Facebook might share content a user has created–especially if required to do so by legislation that mandates interoperability–it is harder to imagine them sharing data they have collected on a user, her tastes and online behaviors. Without this data, likely useful for ad targeting, the large platforms may be at an insurmountable advantage in the contemporary advertising ecosystem.
I explain at length why, even if they had an acceptable user experience, neither the economics nor the technology of decentralized storage networks work in The Four Most Expensive Words in the English Language. First the technology:
Since the key property of a cryptocurrency-based storage service is a lack of trust in the storage providers, Proofs of Space and Time are required. As Bram Cohen has pointed out, this is an extraordinarily difficult problem at the very frontier of research. No viable system has been deployed at scale for long enough for reasonable assurance of its security.
Then the economics:
The second thing that is different now is that the predecessors never faced an entrenched incumbent in their market. Suppose we have a cryptocurrency-based peer-to-peer storage service. Lets call it P2, to emphasize that the following is generic to all cryptocurrency-based storage services.
...
Suppose P2 storage became profitable and started to take business from S3. Amazon's slow AI has an obvious response, it can run P2 peers for itself on the same infrastructure as it runs S3. With its vast economies of scale and extremely low cost of capital, P2-on-S3 would easily capture the bulk of the P2 market. It isn't just that, if successful, the P2 network would become centralized, it is that it would become centralized at Amazon!
Siegele agrees with the MIT Media Lab's report that at present the space of production decentralized applications is minute and most have very few users:
Actual dapps are still few and far between. Graphite, which runs on Blockstack, is a bundle of online word-processor and other office applications, much like Google’s G-Suite. OpenBazaar, which relies on IPFS, is an alternative to Amazon. There is no central server to list what is on offer and to process transactions; instead, buyers and sellers download software that can settle things directly between them. The most popular dapp so far is a game called CryptoKitties, a marketplace for digital pets that lives on the Ethereum blockchain.
So lets talk about CryptoKitties:
Cryptokitties’ popularity exploded in early December and had the Ethereum network gasping for air.
Because the Ethereum blockchain struggled under the load of a simple game:
Ethereum has historically made bold claims that it is able to handle unlimited decentralized applications — to the extent to potentially rival the internet. However, this network disruption has brought caution to accepting that claim. ... The Crypto-Kittie app has shown itself to have the power to place all network processing into congestion.
How many users did it take to cripple the network? It was far fewer than non-blockchain apps can handle with ease:
Citing data from DappRadar, Greylock community lead Chris McCann reports that CryptoKitties has fewer than 1,000 daily active users. Granted, interest has sharply declined over the past few months, but he estimates that even at its peak the DApp likely only had about 14,000 daily users. Neopets, a game to which CryptoKitties is often compared, once had as many as 35 million users.
Total Transaction Fees
What CryptoKitties shows is not that it is possible to build a decentralized game but that it is impossible for such a game to be successful. Current cryptocurrency technology is many orders of magnitude too inefficient for widespread use, and that the price per transaction is only affordable if no-one wants to transact. CryptoKitties average price per transaction spiked 465% between November 28 and December 12 as the game got popular, a major reason why it stopped being popular. The same phenomenon happened during Bitcoin's price spike around the same time.

Early in the report Siegele quotes Hal Varian:
“The only thing more dangerous than an economist is an amateur economist,”
But in this section he writes:
Many Web 3.0 projects have developed their own crypto-economic models. The idea is to replace a centralised firm with a decentralised organisation, held together by incentives created by a token—a kind of “crypto-co-operative”. All those involved, including the users, are meant to have a personal stake in the enterprise and get their fair share of the value created by a protocol.
As, I hope, an especially dangerous amateur economist, I'd like to quote Oscar Wilde, who is reputed to have said:
"The problem with socialism is that it takes up too many evenings"
Stockholders can and do vote at company meetings, but the idea that these votes affect company decisions is a joke. Stock in older companies is mostly held by mutual funds, who reflexively support management. Modern tech companies have equity structures that guarantee that the founders retain control. The same is true of cryptocurrencies, whose HODL-ings are notoriously concentrated in the hands of a few "whales". This inequality is because cryptocurrencies are mechanisms for transferring wealth from later to earlier adopters.

Siegele writes:
The most elaborate working crypto-economic model, however, is Steemit, an online forum which rewards its 1m or so registered users for posting contributions or rating content with real money in the form of steem, another sort of token. One type is liquid and can be cashed out using an exchange, which is meant to provide near-instant gratification and attract users. The other, called “steem power”, is less easily convertible and supposed to keep members engaged: the more they own, the more weight their votes have.
The MIT Media Lab's report points out that this doesn't work:
rewards are not distributed directly in proportion to the raw number of up-votes it receives. The economic success of a post depends greatly on how influential the voters are who vote on their post are, as determined by the amount of SP they hold in the system. Right now, the distribution of SP across users in the system is very unequal -- more than 90% of SP tokens are held by less than 2% of account holders in the system. This immense disparity in voting power complicates Steemit’s narrative around democratized content curation -- it means that a very small number of users are extremely influential and that the vast majority of users’ votes are virtually inconsequential.
In other words, Steemit's governance is centralized for economic reasons.

Paul Saffo said:
Never mistake a clear view for a short distance.
This mistake is endemic in the decentralized Web and blockchain worlds. Siegele seems to have underestimated the distance too, which may be due to the sources he quotes in this section:
  • "Ryan Shea, co-founder of Blockstack"
  • "Joel Monegro of Placeholder VC, a venture-capital firm set up to bet on the trend"
  • "Tim O’Reilly, who pushed peer-to-peer and coined the term “Web 2.0”"
  • "Chris Dixon of Andreessen Horowitz, another VC firm with investments in the field"
Only Tim is not "talking his book". For a set of more independent experts, see Cryptographers on Blockchains: Part 2.

China has the world’s most centralised internet system

The Hamiltonian approach flourishes behind the Great Firewall of China, where with government support the Internet is similarly dominated by a few giants:
Alibaba and Tencent are the acknowledged leaders, particularly in financial services (Baidu, China’s number three, struggles to keep up). With their respective subsidiaries, Alipay and WeChat Pay, they dominate mobile payments. In the big coastal cities, these services have all but replaced cash for smaller purchases and generate immense amounts of data, which the companies then use to target advertisements, improve their e-commerce services and power artificial-intelligence (AI) offerings. Alibaba and Tencent also control much of China’s venture capital. According to McKinsey, a consultancy, between them they make about half of all VC investments in mainland China. In America the tech titans account for only around 5% of such investment.
These large companies don't just "manage" the Web on behalf of the Communist Party, they implement China's "social credit system":
One day in 2017, Hu logged onto a travel site, but couldn’t book a flight because the site said he was “not qualified.” Soon he discovered he was blocked from buying property, using the high-speed train network, or getting a loan. And there was nothing he could do about it. His rights to essential goods and services were now circumscribed through an algorithm designed to discriminate against the 7.5 million people on China's “Dishonest Persons Subject to Enforcement” list.

Welcome to the Chinese “social credit score” system, whose goal is to rank China’s 1.4 billion people. Conceptually, it is not that different from a financial credit score in the US. But the social credit score includes things like political outspokenness, shopping habits, friends, travel habits, and anything the authorities want to encourage or discourage. This score then fine-tunes your access to essential social goods based on a discriminatory algorithm.
Matt Stoller writes:
Such a nightmarish system could never, of course, happen in the United States. Or could it? Three recent decisions in Washington suggest it is not as far-fetched as we might imagine, with both our courts and our government effectively endorsing the way a handful of giant companies are centralizing control over our society.
Note the similarities between the Chinese social credit system and the way Uber and other "gig economy" companies treat their not-employees.

What, apart from social control, is seen as the upside of the Hamiltonian approach? Siegele writes:
Leading thinkers in China argue that putting government in charge of technology has one big advantage: the state can distribute the fruits of AI, which would otherwise go to the owners of algorithms. Feng Xiang of Tsinghua University, one of China’s most prominent legal scholars, recently warned that “if AI remains under the control of market forces, it will inexorably result in a super-rich oligopoly of data billionaires who reap the wealth created by robots that displace human labour, leaving massive unemployment in their wake.” If government can ensure that AI serves society instead of private capitalists, he argues, the technology promises to create wealth for all.
Source
That's unlikely to happen under a government that has created the second-highest Gini coefficient among large economies:
Even if Chinese income inequality is falling, it remains high relative to other big countries—only Brazil has a higher Gini coefficient among the world’s 10 largest economies. The differences in living standards between coastal and inland residents in China remains stark. For the country’s rapid economic growth to become more inclusive, the apparent start of a trend towards falling inequality will need to continue for many decades to come.
As the lucky homeowners of Palo Alto, Vancouver, and other favored cities know, the Gini coefficient:
may understate the level of inequality—but not the overall trend—because wealthy Chinese have become increasingly likely to hide their income from officials (paywall). To construct their series of inequality, the researchers had to combine different data sets based on nationally representative surveys, all of which were collected with some involvement from China’s National Bureau of Statistics.
Note also that the above is about income not wealth inequality.

How regulators can prevent excessive concentration online

It is hard to disagree with Siegele when he writes:
When it comes to democracy and human rights, a Jeffersonian internet is clearly a safer choice. With Web 3.0 still in its infancy, the West at least will need to find other ways to rein in the online giants. The obvious alternative is regulation.
The Hamiltonian approach includes blocking mergers and acquisitions:
When Facebook took over WhatsApp, a popular messaging service, for about $23bn in 2014, the deal barely raised any eyebrows in antitrust quarters. Today the acquisition would probably be blocked, because it has since become clear that Facebook was taking over a serious rival. And despite promises to the contrary, the social network proceeded to merge some of the two firms’ data, which last year earned it a fine of €110m ($122m at the time) from European Commission regulators.
On its own, this obviously won't work. The platforms have enormous engineering resources with which to replicate a merger target's product, and vast cash hoards with which to subsidize undercutting it. So it is also necessary to block some service offerings from big platforms:
because they might favour them over rival offerings. Such a conflict of interest was at the core of an antitrust case in Brussels in which Google was accused of having discriminated against competing comparison-shopping services and fined €2.4bn. Amazon, too, often competes with merchants that use its online marketplace. To avoid such conflicts, limitations have been imposed in other industries, such as railways and banking, points out Lina Khan of the Open Markets Institute. Why, she argues, should this not be possible for platforms?
The Jeffersonian approach, on the other hand:
is variously labelled “data sharing”, “data portability” and, in geekish, “regulation by API” (application programming interface). The champions of data monopolies accept that they will be hard to avoid and even harder to take apart, so they want incumbents to be required to give startups access to some of their data and thus create more competition.
On its own, this obviously won't work. Because the startups are operating on the incumbent's platform, they are extremely well-informed about the startup's business. The moment the startup looks like a potential threat the incumbent can use a tiny part of the massive cash hoard they have accumulated from monopoly rents to acquire and neutralize the threat.

Source
A month ago in American tech giants are making life tough for startups The Economist discussed this problem:
Venture capitalists, such as Albert Wenger of Union Square Ventures, who was an early investor in Twitter, now talk of a “kill-zone” around the giants. Once a young firm enters, it can be extremely difficult to survive. Tech giants try to squash startups by copying them, or they pay to scoop them up early to eliminate a threat.
The article provided many examples, such as:
Snap is the most prominent example; after Snap rebuffed Facebook’s attempts to buy the firm in 2013, for $3bn, Facebook cloned many of its successful features and has put a damper on its growth. A less known example is Life on Air, which launched Meerkat, a live video-streaming app, in 2015. It was obliterated when Twitter acquired and promoted a competing app, Periscope. Life on Air shut Meerkat down and launched a different app, called Houseparty, which offered group video chats. This briefly gained prominence, but was then copied by Facebook, seizing users and attention away from the startup.
It noted that the giants can buy startups very cheaply:
And when startups are bullied into selling, as some are, it is even more worrying. Big tech firms have been known to intimidate startups into agreeing to a sale, saying that they will launch a competing service and put the startup out of business unless they agree to a deal, says one person who was in charge of these negotiations at a big software firm (which uses such tactics).
The result is that startups have to plan to be acquired, limiting their ambition:
Entrepreneurs are “thinking much earlier about which consolidator is going to buy them”, says Larry Chu of Goodwin Procter, a law firm. The tech giants have been avid acquirers: Alphabet, Amazon, Apple, Facebook and Microsoft spent a combined $31.6bn on acquisitions in 2017. This has led some startups to be less ambitious. “Ninety per cent of the startups I see are built for sale, not for scale,” says Ajay Royan of Mithril Capital, which invests in tech.
Startups are not going to achieve a decentralized Web in this environment.

There is no single solution to making the internet more decentralised

Antitrust actions of the past have been successful, as Siegele describes:
Looking back, forcing the tech giants of the past to share some of their wealth seems to have been a good idea. Intel would have found it harder to develop microprocessors without a consent decree in 1956 that forced AT&T, then America’s telephone monopoly, to agree to license all its past patents free of charge, including the ones for the transistor. Microsoft might never have come to rule PC software if IBM, accused of monopolising mainframes, had not decided in 1969 to market computers and their programs separately, a move that created the software industry. Google might not have taken off in the way it did had Microsoft not agreed, at the end of its antitrust trials in America and Europe in the 2000s, not to discriminate against rival browsers and to license technical information which allows other operating systems to work easily with Windows.
But even if the Chicago school ideology would permit it, neither the Hamiltonian nor the Jeffersonian approach will work on the their own. Siegele's ideas for a combination aren't persuasive, either. For example:
Equally important, governments must make it easier for decentralised alternatives to emerge. That could mean creating demand for such offerings either by using them themselves or by mandating their use, for instance by requiring that some of them, such as blockchain-based digital identities, are offered by big online-service providers. But it also means doing away with regulation that ends up strengthening existing online giants.
Governments don't want decentralised alternatives. A few dominant companies are much easier for government to deal with than a plethora of "decentralised alternatives", as we see in China and also in the US, with the cozy relationship between AT&T and the NSA. Siegele suggestions include these:
In America the Computer Fraud and Abuse act (sic) and Digital Millennium Copyright (sic) makes it an offence, punishable by prison, for outside firms to plug into the platforms of online giants. Such legislation should be dispensed with. It is also unhelpful to treat all crypto-tokens as securities and regulate them as such, as America’s Securities and Exchange Commission seems set to do. Exceptions should be made for those that are clearly intended to power new types of services. The European Union may need to tweak its brand-new General Data Protection Regulation (GDPR) to make it less complex. Big firms have the resources to comply with its rules, whereas smaller outfits are likely to struggle.
Where to start with this?
  • Clearly, the CFAA and the DMCA have had disastrous consequences, but a Congress that just passed FOSTA/SESTA is obviously not going to fix them.
  • Regulating crypto-tokens as securities is the least that can be done to address the massive crime wave they have sparked. Even an ICO as "clearly intended to power new types of service" as Filecoin should never have been permitted; it is easy for even an amateur economist to show that investors other than the promoters and a favored few cannot expect to profit.
  • Tweaking the GDPR to allow blockchain-based systems would be to give garbage-in garbage-out the force of law. Even if you believe blockchains are immutable, that doesn't mean the data is right, it just means you can't change it even if it is wrong.
Overall, Siegele's report is a well-written, accessible overview of the problem and of some possible components of a solution. I urge you to read it. But his analysis of the components lacks skepticism. Since a viable solution has escaped everyone else it would be very unfair to criticize him for failing to propose one, but it does make the report rather depressing reading.

Update July 4th 2018

Two quick updates. First, Dan McCrum's Intel's disruption, and the problem with every token pitch at the FT's Alphaville makes the point that proprietary lock-in (centralization) is central to the pitch for almost every ICO:
The essential pitch of most initial coin offerings is a request for people to lock themselves into a new proprietary network. See Iotas, Eos, Equis, Wallos, Owns, Ton$, just to highlight a few of the moles we've whacked.

We've often described such propositions as tokens in search of problems, but maybe its better to think of them as would-be monopolists hoping to sell entrapment. One coin to rule them all, etc.
Source
Second, to give you some idea of how hard, in the real world, just the technological problems of decentralized systems can be, David Gerard:
found out the name of the routing problem the Lightning Network has to solve, to get money from arbitrary person A to arbitrary person B across a mesh of unknown connections of varying usability (liquidity) — the Canadian Traveller Problem. Finding the best path is PSPACE-complete.
So, as I described in Techno-Hype Part 2.5, the Lightning developers were right to punt on the routing problem. But punting means the Lightning network will definitely be centralized around a small number of "banks". Here, from Diar's Lightning Strikes, But Select Hubs Dominate Network Funds, is a table showing that the predicted centralization of the Lightning Network is already well advanced:

Top 10 Lightning Nodes Hold Over 50% of Current Funds

* Private Node - Data Reflects Network on June 23 2018

12 comments:

David. said...

“I Was Devastated”: Tim Berners-Lee, the Man Who Created the World Wide Web, Has Some Regrets by Katrina Booker at Vanity Fair is hagiography about Sir Tim Berners-Lee and Solid, perpetuating the myth that decentralizing the Web is about technology, which is necessary but not sufficient.

Geoff said...

There's a book to be written here, David. You've got a lot of the central elements in your last year's blogging.

David. said...

Andrea James llustrates the "kill zone" for startups in These nine companies are snapping up almost all the AI startups:

"In the current acquisition binge around artificial intelligence, tech behemoths with deep pockets lead the way, including Google, Apple, Facebook, Amazon, Intel, Microsoft, Twitter, and Salesforce."

David. said...

As frequently happens, Diar discovered that writing a factual article that is less than flattering to a cryptocurrency network results in vituperative but fact-free attacks from the developers.

Ludwig said...

Many thanks for your thoughtful comments on my special report. This must be the first time that someone has called me not skeptical enough…;-) More seriously, I think the problem with this assessment is that you misinterpret the report’s descriptive parts as endorsements. You also quote somewhat selectively: the practical problems blockchain projects have to overcome, the fact that they are mostly centralised themselves, the growing “kill zone” for startups—all that is mentioned in the report. Anyway, I don’t think we really disagree on any of this. Where we differ, is on what conclusions should be drawn. Decentralised applications are certainly hard to pull off, but does that mean Web 3.0 startups should stop trying? Only a lot of experimentation can lead to the breakthrough that might cut the “blockchain’s Gordian knot”, in Tim O’Reilly’s words. Yes, skepticism is in order, but also encouragement.

David. said...

Universities are cashing in on the blockchain. All that's needed to collct the cash is uncritical recycling of the hype. Some illustrative quotes:

The London School of Economics: Understand how the distributed ledger technology behind cryptocurrencies will radically reshape the business world and transform major industries

Imperial College: The wider use of cryptocurrencies is the next natural step in reducing friction in the global economy, supported by the adoption of tokens in local contexts, be they specific to geographies or industry-sectors.

Stanford: “Blockchains will become increasingly critical to doing business globally,” said Boneh, the Rajeev Motwani Professor in the School of Engineering

David. said...

"A cryptotoken named ifishyunyu has created congestion on the Ethereum network. This has caused the transaction costs or Gas prices to increase to about 100 Gwei, almost 6x the normal. While Ethereum network has faced a similar situation in the past due to Cryptokitties game, the difference this time is that the cryptocoin is unknown and does not serve any purpose. So, it means that there is deliberate attempt to keep the network slow, leading to frustration and migration to other networks." reports Kelly Cromley in The Cryptotoken Behind Ethereum Network Congestion.

Someone noticed the Cryptokitties spike in transaction costs and decided to DDOS the network.

"This way of creating traffic congestion has exposed another issue that a decentralized, smart contract platform can face at any time. It has shown a way by which a user having thousands of Eth tokens can slow down the network for several months. The issue has once again confirms that blockchain technology has a long way to go before these issues are permanently removed."

David. said...

"Startups are often in position to lead the antitrust charge against major competitors. But entrepreneurs face a dilemma: If they go running to regulators, they have to admit they’re in danger and tick off a powerful player in their world. If they do nothing, they risk bleeding out." from David McCabe's Why startups aren't pushing the feds to break up Big Tech.

He has a good summary of the history of startups pushing anti-trust actions (MCI vs. AT&T, Netscape vs. Microsoft), and of the reasons why these days it doesn't happen.

David. said...

"Last week, we criticised a bad-quality joint report put out by “Imperial College London” and eToro, the crypto-shilling trading firm. The report conveniently supported eToro's conviction “that cryptocurrencies will gain global mainstream adoption within the next decade” and was widely picked up by news sites.

We said the report did not live up to the standards we expect from a university ranked eight-best in the world, for instance by simply attributing blind quotes to a “renowned Bitcoin expert”.

Well, it turns out the report was not the work of Imperial College London after all. A correction reveals it had been mis-badged.

The cryptopuff was instead the work of two academics, facilitated by Imperial Consultants -- a separate commercial entity from the university, though wholly owned by the college."

From Jemima Kelley's Imperial College corrects eToro cryptopuff: it wasn't 'research' after all. A distinction without much difference.

Blissex2 said...

Following another chain of thought I came to a related conclusion: that centralization means that the "end to end principle", of a dumb network and smart endnodes, is dead dead dead.

The "end to end" principle applies to communication flows that are many-to-many, and the centralized internet is instead largely many-to-one: all Facebook users communicate via Facebook, all users of Amazon-hosted services communicate via Amazon.

In effect Google, Facebook, Amazon are pretty much the same as the old phone monopolies: dumb terminals and smart networks. Today the dumb terminals are called "smartphones", but all they do is to connect to a few central highly sophisticated server fleets, just as analog phones connected to sophisticated switches.

HTTP (or rather SOAP and successors) has become the new ISDN.

David. said...

Dave Michel's Here’s how GDPR and the blockchain can coexist is a good overview, less optimistic than the title. As regards personal data on a permissionless blockchain:

"Blockchain data are not really immutable – they’re just hard to change. Collectively, the nodes control all copies of the blockchain. They can change the data stored on the chain by moving to a new version, called ‘forking.’"

So every deletion would be a hard fork, requiring some means of enforcing miners to abandon and delete the chain with un-deleted data. Not going to happen.

Permissioned blockchains could provide the necessary enforcement:

"Admittedly, one could argue that this defeats the point of using blockchain in the first place. If the [nodes] can change the data on the chain, then outsiders can no longer independently verify the blockchain’s integrity.

Whether this matters will differ per application. In some cases, it should be enough that the nodes can verify and vouch for the integrity of the blockchain (since they must agree to and implement any forks). Outsiders may simply trust private blockchain operators to maintain the ledger."

All that mechanism to do what a simple database would do.

But wait! Suppose you just stored the hash of the personal data on the blockchain? One could:

"could take a hash of the [personal data] they wish to verify, by inputting the [personal data] into a hash function. Then, they store the resulting hash on the blockchain, while storing the [personal data] itself off-chain.

Deleting the personal data from the off-chain storage would leave only the hash on the chain. Since hashing is one way, there is no way to discern the original information from the hash."

But if in some way you already have a copy of the personal data, you can identify it with the hash, so that is very fragile. You could:

"‘peppering’ the hash, and makes it much harder for an outsider to link the on-chain hash to the individual.

However, some cryptographers warn that peppered hashes are not fully secure, since they rely on keeping a “server-side secret.” If the security of the nonce is compromised, then outsiders can link the on-chain hash to the individual. At present, it’s unclear whether a peppered hash would qualify as personal data."

And the server-side secrets make this scheme fragile too. Other than the frantic search for viable use cases for blockchain technologies, there's no reason not to just uses a database. Its way cheaper and easier to manager than a permissioned blockchain, and it is no more fragile.

David. said...

"So while merchants can and do accept ETH (and its derivatives) for payment, perhaps a more accurate measure of its activity is how many Dapp users there are.

There are a couple sites that estimate Daily Active Users:

* State of the Dapps currently estimates that there are 8.93k users and 8.25K ETH moving through Dapps

* DappRadar estimates a similar number, around 8.37k users and 8.57K ETH moving through Dapps

From Tim Swanson's How much electricity is consumed by Bitcoin, Bitcoin Cash, Ethereum, Litecoin, and Monero?. About 8,000 users and $2.5K in daily transactions for Ethereum's Dapps. That's a long way from decentralizing the Web!