On problems with social science, VC investing, collusion, and weird physics

Interesting Links: September 13, 2020

…diving into the sea of trash that is social science gives you a more tangible perspective, a more visceral revulsion, and perhaps even a sense of Lovecraftian awe at the sheer magnitude of it all: a vast landfill—a great agglomeration of garbage extending as far as the eye can see, effluvious waves crashing and throwing up a foul foam of p=0.049 papers.

Alvaro Menard, What’s Wrong with Social Science and How to Fix It

1. What’s wrong with social science and how to fix it

   Great post analysing and synthesizing many of the problems in the social sciences. What’s funny is that the author finds it abominable that even the top economics journals barely manage a 2/3rds expected replication rate, while my first reaction was ‘wow, that’s much better than I would have expected!’. The results are much, much worse in most of the other social sciences. Education does quite well (relative to the others) which was refreshing to see. Unsurprisingly, marketing is at the very bottom of the list. And based on my experience, even a 40% expected replication rate is shockingly high - try 10% or any other adequately low made-up number.

   What is perhaps slightly surprising is that even after the ‘replication crisis’ there doesn’t appear to be any improvement. Which means that the institutions and most people inolved (even if unwittingly) don’t actually want to strive for better accuracy and the truth. Ok, maybe this is not surprising at all.

   “Studies that replicate are cited at the same rate as studies that do not. Publishing your own weak papers is one thing, but citing other people’s weak papers?”

   “I think the most plausible explanation is that scientists don’t read the papers they cite, which I suppose involves both malice and stupidity.”

   Robin Hanson has some interesting views on this. I agree more with Robin Hanson on the issues with the author’s stated solution.

2. Bessemer Venture Partners have made available a few investment memos of their successful investments with some current commentary about them. It’s obviously more of a marketing tool and doesn’t cover their memos in unsuccessful investments or about investments they did not end up making but which ended up being very successful. But it nonetheless makes for an interesting read.

   The Shopify memo. It understandably pats itself on the back because back then “very few VCs believed that software businesses targeting SMBs were good investments, let alone ones that could deliver venture returns.” That VCs did not think SMBs are worth targeting is the most predictable and depressing idea. The algorithm for making money is easy - find rich people, and give them what they want (which is invariably more money and power). There is a version of this permeating the recent Indian tech industry as well, when you have the founder of CRED proudly proclaim things like “…it’s insanely hard to monetise people who do not have money” and “a lot of these factors are just completely ignored. The question is there are 30-40 million customers…if you do not build good products for them, why go and chase these 500 million customers, which currently don’t even have money.”

   The Pinterest memo. The way VCs talk is fascinating to me: “We leveraged our relationship with Jimmy Wales (one of the founder’s “Internet heroes”) to grab their attention, but we are unable to mitigate the fact that they have received several term sheets. As a result, this is a very expensive deal, but we think the magnitude of the opportunity justifies the price.” The obsessive use of the word “leverage” by finance professionals and management consultants is telling when a simpler “use” would fit better but sounds a whole lot worse. Oh, and “unable to mitigate the fact” - business writing at its finest.

   Their probabilities of failure (0% return) and 1x return for each of the business was interesting:

   Company	Round	Year	Failure %	~1x %
   LinkedIn	Series C	2006	0.5%	65%
   PagerDuty	Series B	2014	5%	35%
   Twitch	Series B	2012	15%	20%
   SendGrid	Series B	2011	10%	35%
   Mindbody	Series B?	2010	15%	20%
   Pinterest	Series A	2011	10%	35%
   Fiverr	Series A	2011	30%	30%
   Shopify	Series A	2010	12%	23%
   LifeLock	Series A	2006	30%	25%
   Yelp	Series A	2005	25%	38%
   Wix	Seed	2007	29%	30%
   Twilio	Seed	2009	NA	NA
   Dropcam	Seed	2009	NA	NA

3. Vitalik Buterin on the perils of unbalanced coordination.

   Evaluate intentions not actions: He writes “…actions that a person takes are a combination of that person’s internal knowledge, goals and preferences together with externally imposed incentives on that person, and so the actions that people take when colluding, versus the actions that people take on their own volition (or coordinating in benign ways) often overlap.” This is indeed the classic ‘defense’ used in alleged business collusion cases.

   On decentralization: “One naive view of why decentralization is valuable is that it’s about reducing risk from single points of technical failure. In traditional “enterprise” distributed systems, this is often actually true…” But we should look at decentralization as being the presence of barriers to harmful collusion.

   On forking as counter-coordination: Often to outsiders, the presence of so many forks of bitcoin like digital currencies is troublesome. However, as Vitalik points out: “If a system gets taken over by a harmful coalition, the dissidents can come together and create an alternative version of the system, which has (mostly) the same rules except that it removes the power of the attacking coalition to control the system.” And this is perhaps one of the reasons why the vast majority of governments are not particularly big supporters of decentralization.

4. Martin Hairer won the $3 million 2021 Breakthrough Prize in Mathematics for his work on stochastic PDEs aka partial differential equations. He also won the 2014 Fields Medal along side other stalwarts that year like Maryam Mirzakhani (who passed away way too quickly three years ago) and Manjul Bhargava (who was on the Indian National Education Policy 2020 committee).

   This 2014 article in Quanta magazine after he won the Fields Medal is a good, gentle introduction into his work.

5. Update on DARPA’s EmDrive project and Mike McCulloch’s Quantised Inertia theory as an alternative to general relativity.

   “With two ongoing studies rigorously testing the EmDrive’s “impossibility,” the controversial drive that’s hung around astro-engineering circles for more than two decades is only months away from its do-or-die moment.”