Thursday, May 18, 2023

Lies, Damned Lies, & A16Z's Statistics

This post is a quick shout-out to two excellent pieces documenting the corruption of the venture capital industry:
Below the fold I comment on both.

In List And Dump Schemes I discussed Fais Khan's "You Don't Own Web3": A Coinbase Curse and How VCs Sell Crypto to Retail, a description of how by outsourcing securities fraud to cryptocurrency entrepreneurs VCs could turn their money over much more quickly, juicing their returns while avoiding legal liability. This was a big part of the way the flood of money chasing returns in a low interest-rate environment corrupted venture capital.

Molly White

Narrative over numbers: Andreessen Horowitz's State of Crypto report starts by pointing out that A16Z has raised over $7.5B to invest in the cryptosphere, over half of it after cryptocurrency prices started collapsing on 11th November 2021:
$350 million in 2018, $515 million in 2020, $2.2 billion in 2021, and $4.5 billion in 2022.
White documents in detail how A16Z's desperation to show a return on these billions leads to obfuscating, lying, misleading and egregious forms of chart-crime. I'll just discuss three of the worst examples.

The first is from their 17th May, 2022 (almost exactly six months after "prices" started falling) State of Crypto, touting the 270.9% rise in cryptocurrencies' "market cap". White writes:
With the use of some extended x-axes and creative (and undisclosed) data cut-offs, they were able to make it appear that crypto prices could still be on a rocket trajectory. For example, take this chart of “global crypto market cap”. Looks compelling! Up only, which they helpfully illustrate with the crude black arrow in case those pesky “down” portions of the graph were threatening to confuse the message.
To their credit, they do cite their sources — in this case, CoinMarketCap. This allows us to pull the same data from the same source, as of the time at which the report was published:
I have taken the chart through May 2022, overlaid it on a16z’s, and massaged it a little bit to fit, which produces the following (note I’ve turned it into grayscale to help with visibility):
CoinMarketCap data (black rectangle) overlaid on a16z’s chart. The red dashed line marks on the CoinMarketCap chart the claimed cutoff of December 31, 2022.
This allows us to see how a16z cut off the data at roughly May 2021, despite claiming on that slide that the data was as of December 31, 2021 (marked by the red line), and despite in later slides using data through May 2022. This omission conveniently elides the downturns in mid and late 2021 that might make readers notice that “hmm, sometimes crypto prices do go down”.
White doesn't mention her excellent takedown of the whole concept of "market cap" in Cryptocurrency "market caps" and notional value, which expands upon Jemima Kelly's No, bitcoin is not “the ninth-most-valuable asset in the world”.

The second is A16Z's misleading hype of "Web3":
Andreessen Horowitz begins the “Why Web3 Matters” section of the report by repeating a similar refrain as they did in the 2022 report: “web3 is the next evolution of the internet”. They fail to mention that while the columns describing the “web1” and “web2” eras describe actual changes in the web, the “web3” column remains wholly aspirational despite its supposed 2020 start date.

At this point, a critical reader should be wondering why Andreessen Horowitz — a venture capital firm that has backed (and continues to back) some of the largest Big Tech firms, and whose entire business model relies on accruing value to themselves and their investors — would be interested in something that was truly “community-governed” where “value accrues to network participants”.
As Fais Khan shows such value as there might be in "Web3" accrues not to "network participants" but to the VCs themselves via List And Dump Schemes. The idea that A16Z would invest spend $7.5B on something governed by "the community" so that “value accrues to network participants” doesn't pass the laugh test.

In fact this whole 3-era history of the Web that Chris Dixon has been pushing is deliberately distorted to make it look like his vision of "Web3" is inevitable and thus a massive investment opportunity. Dave Karpf's Web3's fake version of Web history supplies a neccessary corrective:
the problem with Dixon’s model is that it extremely, ceaselessly, aggressively wrong. It’s the type of wrong that might be useful for hawking unregistered Web3 security products (err, sorry, I mean, play-to-earn games), but is not at all useful for actually understanding the development of the internet.

1990-2005 wasn’t a single, contiguous era of “open decentralized protocols” and value accruing to the edges of the network. There were (at least) three eras in that timespan. Only the first (1990-95) had those qualities. As soon as the money got big, things changed drastically.

2005-2020 wasn’t a single era either. Again, there were at least three eras in there. And only the last one or two fit his description of “siloed, centralized services” with value accruing to the big tech companies. The years that most clearly represent the “web 2.0” era were characterized by social sharing and mass collaboration. It was only later that the platforms calcified and the “enshittification” cycle began in earnest
Karpf concludes:
But Chris Dixon was there for too much of the history of the Web to be making innocent mistakes here. When he erases Microsoft from the ‘90s Web, it isn’t because he never heard of the browser wars. When he conflates the participatory Web 2.0 years with the platform years that followed, it is an intentional omission. He’s getting the basic history wrong because it serves his strategic purposes as a Web3 investor and evangelist.
You know why Web3 has turned out to be so much scammier than the internet of the 90s and 00s? The answer is simple. It’s the same reason why Willie Sutton robbed banks (“Because that’s where the money is!). You can’t code community participation and trust into the blockchain. Once the money gets big, the social incentives get skewed. In the complete absence of regulation, people are going to run huge scams.

We’ve seen the result. Web3 has been a catastrophe for everyone but the early investors and the scammers. Chris Dixon constructed a model of Web history to help sell his investments.
The third, A16Z's tribute to the reduction in Ethereum's energy consumption from the switch to Proof-of-Stake, is a real doozy. It centers on the claim that it "eliminates environmental objections". White counters:
Given they discuss Bitcoin throughout this report, it seems a little disingenuous to say that environmental concerns have been eliminated. It’s correct that The Merge greatly reduced Ethereum’s energy consumption, which is excellent! But Bitcoin still has a massive carbon footprint, which is currently comparable to that of the entire country of Peru. Its electricity consumption is comparable to that of Kazakhstan. If Bitcoin itself was a country, it would rank 34th in terms of energy consumption.
Well, yes, but as I noted in The Power Of Ethereum's Merge, Ethereum only consumed half of Bitcoin's power, so the Merge only eliminated 33% of the environmental objections. Actually, the decrease was much less because (a) there are many other Proof-of-Work cryptocurrencies apart from Bitcoin, and (b) many of the slots in mining centers previously occupied by Ethereum rigs migrated to holding Bitcoin rigs, increasing Bitcoin's consumption. And note that even Proof-of-Stake Ethereum uses vastly more power (and is much slower) than an equivalent centralized system running on a few Raspberry Pis.

But that isn't the doozy, which is A16Z's comparison of Ethereum's power consumption to that of YouTube, which includes the claim that while "global data centers" consume 200TWh/year, somehow YouTube consumes 244TWh/year! White traces the sources for this amazing claim in detail, showing that no-one ever thought about the text they cut-and-pasted, and concludes:
Google’s own annual environmental report for 2022 shows they used 18.6 TWh across the entire company.21 Only a portion of all of the electricity used by Google can reasonably be attributed to YouTube, but I couldn't find more granular figures. Either way, a16z's estimated 244 TWh is off by more than an order of magnitude, and possibly even closer to two orders of magnitude — something that should have been apparent at a glance to whoever wrote the report.
I hope these samples encourage you to go read the whole of Molly's epic takedown.

Matt Levine

The aftermath of the global financial crisis led to an extended period of very low interest rates. This created a flood of money chasing higher returns, and this in turn created huge demand for the supposedly superior returns from venture capital. The VC's approach to investing this cornucopia is best summarized by Sequoia's "due diligence" before investing in FTX, which consisted of watching Sam Bankman-Fried playing (not very well) League of Legends.

In Crypto Had Its Bank Runs Too, Levine discusses the recent performance of another prominent VC, Tiger Global, quoting the Financial Times:
In 2022, Tiger’s flagship fund suffered its worst annual loss, losing more than 50 per cent of its value as Tiger marked down its unlisted holdings by nearly 20 per cent.
Levine explains:
In the startup boom of recent years, Tiger Global Management got a reputation for investing in every startup, moving fast, paying top dollar and not being too involved in governance.
This worked because most VC investments fail, their returns come from the small fraction of big successes. It is difficult and time-consuming to spot the big winners ahead of time, so Tiger's approach was to be in as many deals as possible. Doing "due diligence" on the deals was thus not merely a waste of time, but actively counter-productive because it would alienate the founders.

That was during the boom, but after the boom comes the slump:
Technology-focused hedge fund Tiger Global is exploring options to cash in a piece of its more than $40bn portfolio of privately held companies, according to people familiar with the matter.
Levine points out the problem with this:
If you go to the secondary market with a bunch of startup stakes and a reputation for paying top dollar and not doing much due diligence, people are going to want a discount.
You should read the whole Tiger Global section of Levine's post, if only for his 7-point breakdown of why the Tiger Global "founder friendly" approach was justified in a boom. I'll return to the rest of Levine's post in a subsequent post here.

1 comment:

  1. Molly White returns to critiquing Chris Dixon in Review: Chris Dixon's Read Write Own and it is a must-read:

    "Anyway, fear not, says Dixon, because he has found the solution to the internet's Big Tech sickness: blockchains. "While plenty of people recognize their potential—including me—much of the establishment disregards them," complains a general partner at one of the most powerful venture capital firms in the web space. Now, if we would all just be so kind as to ignore the last fifteen years since blockchains' inception — during which innumerable companies have flailed around trying to find any possible use case beyond the manic speculation that has enriched a few at the expense of many — he's got an idea to sell us."