{"id":16955,"date":"2020-08-08T21:09:33","date_gmt":"2020-08-08T21:09:33","guid":{"rendered":"https:\/\/davidgerard.co.uk\/blockchain\/?p=16955"},"modified":"2025-12-31T00:15:16","modified_gmt":"2025-12-31T00:15:16","slug":"balaji-srinivasans-plan-to-save-journalism","status":"publish","type":"post","link":"https:\/\/davidgerard.co.uk\/blockchain\/2020\/08\/08\/balaji-srinivasans-plan-to-save-journalism\/","title":{"rendered":"Balaji Srinivasan’s plan to save journalism"},"content":{"rendered":"

With the Slate Star Codex blog’s self-inflicted self-own<\/a> as a casus belli<\/i>, Silicon Valley is coming out in force against vexatious turbulent journalists who will surely discredit journalism itself<\/i> by … reporting on venture capitalists’ abuses of their employees. [Vice<\/i><\/a>; <\/i>Verge<\/i><\/a>]<\/p>\n

The charge is being led by Bitcoin startup disaster area Balaji S. Srinivasan \u2014 who has, in the last couple of months, been tweeting a plan to replace the entire profession of so-called “journalists” who have the temerity to question him and his VC pals.<\/p>\n

Srinivasan’s solution: paid voting on factual claims \u2014 on a blockchain!<\/p>\n

 <\/p>\n

\"\"<\/a><\/p>\n

 <\/p>\n

The crimes of journalism against venture capital<\/h3>\n

In December 2019, The Verge published a report on an abusive employer \u2014 who was funded by the Silicon Valley VC crew. [Verge<\/i><\/a>] A journalist then tweeted in June 2020 about the abusive employer in question. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p>\n

Srinivasan claimed this is unfair when the abusive employer is … female? Apparently abusing employees is fine if it’s a woman doing it. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>] Srinivasan then called down a social media storm upon the second journalist \u2014 he considered this a perfect example of how journalism was corrupt \u2014 and started talking about how to replace these people.<\/p>\n

The blog Slate Star Codex shut down for a month in late June \u2014 and that was the last straw for the VCs. The New York Times was writing a profile of this incredibly popular blog, and was going to use the founder’s \u2014 openly publicised<\/a> \u2014 real name in the article about him, and not just his pen name. The author claimed the NYT was “threatening” him and would “doxx” him, so he closed the blog and called for his readers to contact the NYT. [Slate Star Codex<\/i><\/a>, <\/i>archive<\/i><\/a>] SSC’s venture capital readers reacted with outrage, and treated SSC shutting itself down as part of the same issue as the Verge story existing. [New Yorker<\/i><\/a>]<\/p>\n

Journalism, right, but on the blockchain<\/h3>\n

Srinivasan’s plan to fix journalism is … automated voting on the truthfulness of new facts! With “a button that allows the public to pay crypto to crowdfund more confirmations from specific parties.” Replace editors with a prediction market.<\/p>\n

Srinivasan has not presented this publicly as a written plan. He did, however, tweet a sketch of his idea at length in early July, and answered questions in more detail. So he’s clearly got a worked-out picture in his head.<\/p>\n

Here are the tweets (that I could find), with some explanations of the more condensed jargon:<\/p>\n

The truefeed<\/p>\n

Looks like a newsfeed, except the most visually prominent features are:<\/p>\n

1) the number of independent confirmations by economically disaligned parties, with links
\n2) a button that allows the public to pay crypto to crowdfund more confirmations from specific parties<\/p>\n

What goes to the top of the feed?<\/p>\n

Not just the most popular stories, the ones with the most independent confirmations. With the most economic activity invested in proving them true or false.<\/p>\n

The very design of the feed would encourage confirmation rather than mere repetition. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p>\n

(question: Who decides what counts as a “confirmation?” What’s the penalty (and who decides) if a confirmed story turns out to be false?)<\/i><\/p>\n

1) The people crowdfunding it determine whether it is an acceptable confirmation for release of funds. Similar to editors looking at a reporter\u2019s story or VCs evaluating an associate\u2019s diligence.
\n2) This is tricky because things claimed to be false at time X can become true.<\/p>\n

However, a profile page for each user that shows their past confirmations vs current consensus* and other users could be helpful.<\/p>\n

Here\u2019s how we could determine if confirmers are \u201cdisaligned\u201d or not. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p>\n

\"\"<\/a><\/p>\n

(question: More specifically, let’s say the WSJ just reported that Uber is buying Postmates. Does a confirmation consist of me, independent journalist, saying “my sources tell me this is true?” Do I have to detail who my sources are? if it’s the former this seems trivially gamable (I just “confirm” everything the NYT and WSJ reports). If it’s the latter you’re not going to get takers for anything that depends on anonymous sourcing\u2014which is a lot of stories that matter.)<\/i><\/p>\n

The use of truly anonymous sources is a debit from your trust bank. However, pseudonymous cryptographic attestation via zero knowledge is an alternative.<\/p>\n

For example, prove you are Satoshi by signing a message but don\u2019t reveal anything else.<\/p>\n

See here: SFBW19: The Pseudonymous Economy [YouTube<\/i><\/a>] [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p><\/blockquote>\n

Zero knowledge proof (ZKP):<\/i> a method of checking an encrypted claim, without decrypting the claim. The blockchain world has long been enthusiastic about the possibilities of zero knowledge proofs, and talks about them as if you can just assume they work in practice \u2014 though there are, as yet, no production examples in blockchain. Srinivasan’s video talks like ZKPs might as well be here already \u2014 even though they still aren’t.<\/p>\n

(response: You should build this.)<\/i><\/p>\n

Glad to find us aligned on something. This could help stop the spread of misinformation on things like COVID.<\/p>\n

It would also financially incentivize people with expertise to arbitrate contentious technical issues (eg HCQ). [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p><\/blockquote>\n

HCQ:<\/i> hydroxychloroquine \u2014 an existing drug for lupus that was proposed as a treatment for COVID-19, and aggressively promoted by President Trump, but which didn’t work out in clinical trials. Hydroxychloroquine is now a favourite of medical cranks.<\/p>\n

As oracles come online, we’ll get an open source Reuters equivalent. For each fact we know:<\/p>\n

1) Who (digital signature)
\n2) …said what (hash of content)
\n3) …and when (timestamp)<\/p>\n

These facts drive financial decisions so will be checked aggressively.<\/p>\n

Blockchain Oracles Explained [Binance Academy<\/i><\/a>] [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p><\/blockquote>\n

Oracle:<\/i> a machine-readable source of data, considered trustworthy. You need an oracle for a blockchain-based smart contract to know anything about the world outside its blockchain. Srinivasan’s plan seems to involve people paying to vote on journalistic claims, and using that as an oracle.<\/p>\n

The ledger of record is the set of all cryptographically signed feeds of on-chain data. It subsumes social media feeds, data APIs, event streams, newsletters, RSS. It’ll take years to build, but will ultimately become the decentralized layer of facts that underpins all narrative.<\/p>\n

Think of the ledger of record as a decentralized wire service. Every person & organization slowly moves from posting to centralized social media platforms to posting to decentralized protocols. The latter have monetization, permissions, distribution, and programmability built in. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p><\/blockquote>\n

A number of immediately obvious issues with this plan<\/h3>\n

I’m sure the tweets I found don’t fully explain Srinivasan’s plan as he would like it covered \u2014 so I look forward to him setting it out properly in detail. But the ideas behind the tweeted sketch of the plan are clear, and the statements made so far need addressing.<\/p>\n

The goal here is the Silicon Valley dream of automating away the human element.<\/p>\n

Prediction market ICO Augur tried to replace human judgement with a betting market \u2014 to serve as an oracle for all manner of claims!<\/p>\n

Augur fell to (1) querulous rules-lawyering<\/a> over ambiguous claims that went to six rounds of disputes over the outcome \u2014 a common issue in existing centralised prediction markets, let alone real life \u2014 and (2) not many people bothering to show up<\/a> and vote on all this stuff.<\/p>\n

I’m pretty sure Srinivasan thinks the bit where readers pay to vote is all that’s needed to correctly line up the incentives, so you will surely make more money dropping quality truth than pumping out tasty nonsense \u2014 contrary to all previous human experience. And just imagine the stories that readers like Srinivasan’s friend Peter Thiel could afford to vote up.<\/p>\n

Srinivasan seems to view the purpose of journalism as the generation of a ticker-tape of grey facts \u2014 and that this is all you need for “an open source Reuters equivalent.” His “ledger of record” is a child’s version of blockchain dreams, where you put everything in the world<\/i> onto one blockchain to rule them all \u2014 and never mind scaling issues or feasibility, even before we get to whether this is a good idea.<\/p>\n

But the facts in the news are interesting because they\u2019re social<\/i> facts \u2014 they’re statements that mean something in a context. There is no shortage of raw data \u2014 people read news at all because you’re using narrative to explain why these facts are noteworthy.<\/p>\n

You have to put the data into a meaningful context \u2014 why should the reader care? You can’t automate away the question “Yes, but why are you telling me all of this?”<\/p>\n

In real life, journalists spot interesting malfeasance and have to fight like hell to get it past uninterested editors and out to the public.<\/p>\n

I was particularly struck by “These facts drive financial decisions so will be checked aggressively”. I’m not so sure we could replace Dan McCrum with a prediction market when we need to call out the next Wirecard.<\/a> The Wirecard story took five years to take down a blatant billion-euro fraud.<\/p>\n

And McCrum relied utterly on the backing of the Financial Times, a strong paper with strong opinions on frauds in the market \u2014 particularly when the “checked aggressively” was private investigators following him, and the German regulator trying to prosecute him for looking too deeply into the fraud.<\/p>\n

But I’m also not so sure Srinivasan would think taking down the Wirecard fraud is the sort of thing journalists should even be doing \u2014 after all, what if they looked a bit closely at the Silicon Valley ethos of “fake it till you make it”? Which, if you’re taking money from investors, is a phrase meaning “knowing material misstatements” \u2014 also known as “fraud.”<\/p>\n

Dr. Srinivasan is a smart dumb guy, not a dumb dumb guy. But this is idiot nerd nonsense that’s been tried repeatedly \u2014 it’s a stupid ICO idea, that multiple ICOs tried variations of. This is the particular species of dumbass that a crypto guy comes up with.<\/p>\n

I ran through a few of these<\/a> a couple of years ago \u2014 Po.et wanted to register journalistic works on their blockchain, and pay journalists in their tokens. Civil had an incomprehensible voting-based governance structure,<\/a> and wanted to pay journalists in their tokens. PressCoin would have users vote on stories, paying to do so in tokens.<\/p>\n

This stupid ICO idea was tried repeatedly, and always cratered. Srinivasan has no excuse not to know better. I mean, at least he didn’t start an ICO for Ledger of Record tokens.<\/p>\n

But it gets even better \u2014 Srinivasan thinks paid voting on scientific facts will work out well. Voting on COVID-19 treatments, no less \u2014 he wants random Silicon Valley nerds to vote on medical questions, and whether quack treatments work. Alex Jones, right, but on the blockchain.<\/p>\n

Srinivasan is famous for his previous attempts to automate away the human element in actual health concerns \u2014 in 2014, he proposed to replace the US Food and Drug Administration with “a Yelp for drugs”. In 2017, he met with President Trump to pitch himself for head of the FDA. [Vox<\/i><\/a>, 2017; <\/i>StatNews<\/i><\/a>, 2017<\/i>]<\/p>\n

Just before Srinivasan went for the FDA job, he deleted all his tweets up to that time \u2014 which is why I’ve taken care to archive the present selection. If only we’d had the ledger of record in 2017!<\/p>\n

You probably don’t remember Srinivasan’s previous foray into journalism \u2014 “Nakamoto”, a crypto news site, from way back in the obscure and little-known past of … January 2020. He lost interest after a very short time, and nobody’s added anything to the site in the past seven months. [Nakamoto<\/i><\/a>; <\/i>Modern Consensus<\/i><\/a>]<\/p>\n

OK, smart guy, what would you<\/i> do to fix journalism?<\/h3>\n

To fix journalism, fund it \u2014 that’s the big problem. There’s a lack of funding for deep dives, and an obvious incentive to clickbait. And if I had a solution to funding journalism, I’d be putting it into practice.<\/p>\n

I suspect directly asking readers for money is the closest we have to an answer at present. Paywalls work great for the financial press \u2014 because their readers have money \u2014 and the Guardian does surprisingly well with no paywall, and has over a million readers happy to give them three quid a month despite that. [Digiday<\/i><\/a>] And even the Guardian just laid off a pile of staff.<\/p>\n

Direct sponsorship works a bit \u2014 and Srinivasan is very on board with the idea of journalists striking out independently. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p>\n

Consider this article you’re reading now as Srinivasan’s idea in practice \u2014 citizen journalism, funded by my readers. Send me $5 or $20 a month, you can probably expense it.<\/a><\/p>\n

Srinivasan also suggests an RSS feed reader \u2014 remember Google Reader? \u2014 but using crypto, for micropayments! So it’ll suck. [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>]<\/p>\n

Srinivasan gives us a vision of paid clickbait \u2014 just directly-funded.<\/p>\n

The other thing we need is protections for journalism \u2014 the “billionaire problem,” when someone has enormous resources to bring to bear against coverage they don’t like, and sue it out of existence, or just buy it and shut it down.<\/p>\n

Look, I’ve been at the sharp end of journalism in action \u2014 I used to be the UK volunteer press guy for Wikipedia, and we got some ridiculous<\/em> goddamn nonsense run about us, including extended attacks on me personally. But that doesn’t make journalism one dot less necessary. Srinivasan doesn’t show awareness of this.<\/p>\n

Whatever the solution is \u2014 it isn’t Srinivasan’s plan.<\/p>\n

But, this guy presumably has people listening to him for reasons. What’s Srinivasan’s track record of achievement?<\/p>\n

 <\/p>\n

\"\"<\/a><\/p>\n

 <\/p>\n

Srinivasan at 21.co \u2014 how to set $125 million on fire, with Bitcoin<\/h3>\n

Srinivasan started with a genuine success \u2014 his genetic screening startup Counsyl was highly profitable, and was eventually sold in 2018 for $375 million. [Press release<\/i><\/a>] Hot off Counsyl’s early success, Srinivasan joined Andreessen Horowitz \u2014 where he brought some serious interest in Bitcoin to the company. [TechCrunch<\/i><\/a>]<\/p>\n

Why do I call Srinivasan a “Bitcoin startup disaster area”? Because of his company 21.co, later Earn.com \u2014 which set an alleged $125 million on fire. Tim Swanson has a research note from 2018 detailing the history of 21.co. [Great Wall of Numbers<\/i><\/a>, 2018, archive<\/a><\/i>]<\/p>\n

If there’s a journalist around, they could probably write quite a good “follow the money” story detailing where that $125 million went \u2014 and into whose pockets.<\/p>\n

21e6 was a Bitcoin mining chip startup. It was started in 2013 by a team of five \u2014 including Srinivasan. Funders included Peter Thiel, and both Marc Andreessen and Ben Horowitz.<\/p>\n

21e6 ran into trouble in 2014 \u2014 when the price of Bitcoin went down, but mining difficulty went up. In 2015, the company rebranded as 21.co, and pivoted from mining hardware to consumer-focused Internet of Things devices \u2014 or, at least, Srinivasan marketed 21.co to prospective funders as doing consumer IoT. [Pitch deck images<\/i><\/a>] Per Swanson:<\/p>\n

One of the 21.co engineers was even interviewed<\/a> on a (now deleted) podcast where he spoke about how drone owners would pay tolls denominated in bitcoin to cut across airspace over yards in your neighborhood. You know, the usual word salad and shower thoughts on social media.<\/p><\/blockquote>\n

The plan was to get 21.co’s inventory of unsaleable and worse-than-useless obsolete Bitcoin mining chips \u2014 and put them into consumer household gadgets. [FT Alphaville<\/i><\/a>, free with login<\/i>]<\/p>\n

On 21 September 2015, 21.co announced its flagship product: the 21 Bitcoin Computer! This was a Raspberry Pi controller \u2014 a tiny $25 single-board computer for kids and hobbyists \u2014 with one of the obsolete 21e6 mining chips attached. A snip at $400!<\/p>\n

The chip’s efficiency was 0.17 joules per gigahash \u2014 you could do better mining bitcoins on an AMD video card of the time.<\/p>\n

I can’t find the original post \u2014 but \/r\/buttcoin quickly christened this device the PiTato. And someone put up the potato image above. [Reddit<\/i><\/a>]<\/p>\n

Sales figures were never disclosed. Swanson says:<\/p>\n

Prior to its discontinuation, 21.co representatives approached multiple well-known Bitcoin developers to help resell the devices. In short, these developers were offered to buy 21.co devices at wholesale prices and expected to resell them at the retail price. It is unclear how many (if any) developers did so.<\/p><\/blockquote>\n

One of Swanson’s commenters notes: “Balaji taught a course at Stanford about cryptocurrency and basically used it to promote 21co and Pitato to students. He gave it for free to students but all the labs were on this hardware.”<\/p>\n

Buoyed by the failure of the PiTato, 21.co pivoted once more \u2014 to Earn.com. Swanson called it “Amazon Mechanical Turk, but with Bitcoin.” You could do mundane tasks and get Earn Tokens for them.<\/p>\n

Earn also had a sideline in charging people to send you email \u2014 this was pitched to VCs who didn’t want their time wasted. In April 2018, Earn started running ICO airdrops \u2014 pushing the promotions to their users’ paid email boxes.<\/p>\n

In April 2018, Earn was merged with Coinbase. This was touted as a fabulously successful startup exit! Though it looked to observers much more as if Andreessen Horowitz had taken a floundering portfolio company (Coinbase), merged it with a disastrous failure of a portfolio company (Earn), and claimed a huge success.<\/p>\n

Srinivasan became Coinbase’s CTO \u2014 though he left after a year. [Cryptoslate<\/i><\/a>]<\/p>\n

Earn.com currently redirects to a Coinbase “educational” page on how to throw your money away on incredibly trashy minor altcoins, many of which appear to be in Andreessen Horowitz’ crypto portfolio. [Coinbase<\/i><\/a>]<\/p>\n

But let’s be clear: Srinivasan did nothing wrong \u2014 as he explains in his blog post, Everything Good About The PiTato Factory Was Me, And Everything Bad Was Those Other Fools, by Balaji Srinivasan, Super Genius. He definitely had no involvement in the company that he founded and was pitching its entire existence, got that? [Medium<\/i><\/a>, 2018, archive<\/a><\/i>]<\/p>\n

Replacing crypto VCs with a very small PiTato<\/h3>\n

GPT-3 is OpenAI’s latest automatic text generator. It’s a new and genuinely better AI-based text generator.<\/p>\n

There’s a demo where you type some text describing a web page and GPT-3 builds the CSS. This is legitimately a pretty cool demo \u2014 but it blew some Silicon Valley venture capitalists’ minds, and they’ve spent the last month proclaiming the robo-apocalypse. [Twitter<\/i><\/a>; <\/i>blog post<\/i><\/a>] Though even Sam Altman from OpenAI, and previously a VC at YCombinator, thinks the hype is a bit much. [Twitter<\/i><\/a>]<\/p>\n

Srinivasan thinks GPT-3 might help replace those pesky journalists \u2014 sports reporting is box scores surrounded by text, right? Let’s GPT-3 the text! [Twitter<\/i><\/a>, <\/i>archive<\/i><\/a>] This is surely a brilliant new idea in 2020, and certainly there’s no need to, e.g.<\/i>, look up automated journalism<\/a> on Wikipedia and see that it’s already been done for a while now \u2014 and is somewhat less useful than just reporting the scores.<\/p>\n

For those of us who can discern nuance when reading text, GPT-3 will be an amazing and amusing text generator that will do all the work of writing a new @dril bot for you. A fully automated word salad tosser.<\/p>\n

Except the human labour of selecting the source texts and screening the output texts \u2014 but we can probably outsource that bit, and never admit it exists.<\/p>\n

Venture capitalists’ fear of GPT-3 is that you could replace their Twitter with a very small shell script, and readers would compliment them on the improvement. Artificial intelligence so advanced, it can make a crypto VC worry that they’re disposable too.<\/p>\n

 <\/p>\n

\n
\n

These VC hating twitter accounts are funny but nothing is as hilarious to me as hearing stories over the past few months from the strange volume of people I've somehow met that have walked me through their work ghostwriting for VCs over the years.<\/p>\n

— Michael Dempsey (@mhdempsey) February 20, 2020<\/a><\/p><\/blockquote>\n