Google Solves Its GSP Information Asymmetry Problem with Conversion Data
… secret deal with MasterCharlatan [tm]
https://www.zerohedge.com/news/2018…etail-purchases-thanks-secret-deal-mastercard
tl;dr – MasterCard is helping Google swindle it's own advertisers and get a bigger commission – in theory, Google can now capture 70% of the real economic profit in the retail part of the economy that uses MasterCard. Game over for the real world.
…
Google's sales algorithm is called a 'Second Price Auction'. There is a theorem, in economics[1], that certain auctions contain perfect information, provided the participants tell the truth about their bids (truth telling in bidding is optimal for VCG auctions). But there is an option open to them, in Generalised Second Price auctions like Google's (GSP) called bid shading where there is an incentive to make 'untruthful bids' (would you believe… I'll pay 10 dollars. Not a penny moar. Oy! The pain!) Is it anti-Semitic to speak of 'Truthful Bids'?
[1]: https://en.wikipedia.org/wiki/Vickrey_auction
There is something Google doesn't (or didn't) know about its ADVERTISERS. it can spy on how many clicks those advertisers get, and it knows what those advertisers spend (including the untruthful bids of the losers in the auction) – but it can't know the truthful bid, which is how much of the CONVERSION REVENUE the advertiser would be willing to give up. Cluestick: on the margin, all of it. If the company's shaded bid was 1.01 USD for a 1.07 USD conversion, then Google can chisel six cents on the dollar out of that advertiser now.
Think of it this way: Google runs an indoor bazaar, like many 'antiques' malls, and charges the sellers in that bazaar rental for their booths. How much can Google charge? They can walk around and see how much traffic is at each booth. It can count those transactions ('clicks' are not quite like listening to ka-chings on the floor, but they are close), and it can see the 'untruthful bids' for booth rentals. What it can't see is the GOLLUM QUESTION… What has it got in its pocketses? What do you have in your wallet. We wants to know.
How much do you want to bet there is a unit at Google that is modelling the balance sheet of its advertisers (both industry typical and big players), doing ratio analysis, and trying to guess how the Sales they can see relate to the advertising spend, what 'industry standard benchmark cost of sales' really are – in order to shake their customers down for EVERY LAST MONOPOLY PENNY?
In the absence of this information – a true blind auction where you don't know everything about the bidders – the best Google can hope for is tweaking their reserve price. This is the EOS result for GSP (after Edelman, B., M. Ostrovsky, M. Schwarz[3]). I'm sure they've spent most of the last decade trying to do this – they are after all a profit maximising business – but the essential information was previously unobtainable. How much can you wring out of each bidder, in the Robinson Caruso / Friday sense, or the Bowley-Edgeworth box, before they decide to take their chips and go home. 'Autonomy vs bargaining'.
[3]: https://www.benedelman.org/publications/gsp-060801.pdf [ PDF!]
Unlike the VCG mechanism, GSP generally does not have an equilibrium in
dominant strategies, and truth-telling is not an equilibrium of GSP.
I assure you, finding out the conversion revenues of their ADVERTISERS (they want to data rape their actual customers, not just you, their victims) is the main source of uncertainty, in what they likely consider their most important metric, 'RPM' (Google's revenue per 1000 impressions). In advertising, it's all about the RPM. ('M' is an old advertising term for 1000, not a million. It comes from Latin mille, as in the old slang for money, C-notes and M-notes).
Google probably regrets choosing second-price (GSP) auctions over VCG auctions and then becoming an industry standard. It's one of those things, like the QWERTY keyboard, you would like to go back in time and fix, but can't.[4]
[4]: http://people.bu.edu/fdc/DGPspring2017.pdf [ PDF!]
…
mpact analysis: this is the same as if Google and Amazon were to complete a merger, with a large proportion of sales data and fulfillment data aggregated in the same company as the front end of the sales funnel (online search). There is some data out there no in Google's grasp yet … maybe. Data about 'local' searches like Yelp and maybe FaceBook – but Google has Android so they can see where you are and probably cheat and look at the messages – and the VISA system. And whatever portion of the economy is 'internal' to Amazon and e-bay/paypal. My guess is that MC retail transactions at POS or online, after a Google search, is the lion's share of the real Economy.
THIS DEAL GIVES GOOGLE THE LAST PIECE TO THE MONOPOLY PUZZLE BOARD. THIS IS BOARDWALK WITH FIVE HOTELS.