Thursday, September 17, 2020

@pdxsag Guest Review of The Analyst by "Y.N."

With the absurdity of the Snowflake IPO today (market cap $60B, >100x sales, Berkshire-Hathaway subscribed for $250M shares at the IPO price), I thought I should take the opportunity to do a review of the novella “The Analyst,” which was written by someone who went by "Y.N." in the CBS comments.

I first read it 2 years ago when it was brought to our attention by the author in the comments to a blog post. It is no less topical today than when it was written.

It is a hard book to review without giving away any spoilers. What I can say is that it combines the existential questions of great science fiction with a financial plot-line CBS readers would definitely appreciate.

Of the questions it brought to mind, one was how would we know whether any of the numbers Netflix, Amazon, FB, or any of these SaaS companies are real? What if 80% of social media is just a bunch of cloud services streaming from each other? What if 80% of Robin Hood trades are bots shuffling fractional shares amongst each other?

With billions at stake, why wouldn't a VC create a bunch of AWS instances that would be real, dues-paying customers to one of the portfolio companies they are prepping for IPO? With no moral compass, and a return of $20 for every $1 in revenue spent at the IPO-ing company, they'd be crazy not to.

As pertains to Robin Hood, what if those thousands of new traders every month are fake accounts made by Robin Hood's VC owners? It wouldn't need to be an inside job. Robin Hood itself could be completely ignorant, albeit willfully and with complete plausible deniably, to be sure. But with a couple hundred dollars and a stolen social security number from the dark web, it shouldn't be hard to open a fake account.

So the question becomes, how would investors on the outside know? What would be the "tell" that a SaaS company were faking its users with bots?

Jeff Bezos would know. The NSA would know. Perhaps a few network architects at Comcast, ATT, and the small number of tier 1 internet service providers might know. Overall, however, not that many people would be have access to the hard data to know. It would be easy to keep such a conspiracy covert. But what about the circumstantial evidence?

One beauty of SaaS fraud versus, say, a car company or telecom equipment manufacturer is the ability to hide the "stuffed channel." It is no easy task hiding 10,000 cars. Obviously they are big and expensive. If you claim to have built and shipped 100,000 cars in a quarter, you can be guaranteed that some meddlesome short-seller, or 30, have staked out your manufacturing plant and counted the cars as they've rolled off the assembly line and out the gate. In contrast, with SaaS fraud, it's just bits on a network connection, and that's assuming they've gone to the trouble of spinning up a user on the outside. As long as money is cheap and price to revenue is greater than 1.0, as a major investor you would continue paying for the service and even be ready to add a customers anytime a quarter looked like it was going to be a little light. Again, with no moral compass, you'd be stupid not to. The SEC certainly isn't going to investigate. Even if they wanted to (don't laugh), how would they? Bits went in and bits went out, no one has any way of determining if the bits had no real purpose.

The next question is how much AI is behind all the social media we interact with on a daily basis? We are used to thinking in terms of algorithms that serve up targeted advertising to us everyday. How much content is algorithmically generated? I assume readers are familiar with internet sock puppets. We think of them as having a human hand inside the sock, but is that necessary? Could a sophisticated AI operate a significant number of sock puppets to drive internet discourse around a topic among real investors? What about a small cabal of 6-12 individuals using AI and bots to force-multiply their actions? The chat bots used for customer service use AI to force-multiply real customer service representatives. How much of our interaction is real and how much is with the bot?

I believe once you start considering that then you have to start considering how much of your own real-life actions are influenced by virtual interactions. We may scoff at Facebook users falling for obvious conspiracy theories meant to discredit the people that believe them, but where is the line between obvious dis-information the “rubes” fall for, and the not so obvious dis-information one personally falls for? How many of your social media “friends” (including bloggers) talking about something in a believable manner would it take to move you to action? Especially in this day when all valuations are relative, what does one use as their check for reality?

I don't have answers. The pleasure of science-fiction is having an author pose interesting questions to the reader which he must ponder the answer to for himself. In this regard, “The Analyst” was excellent.


No comments: