Re:The case for why Google should be regulated as a public u...
by pb7 on Tuesday August 11, 2020 @12:41PM
(#24122357)
>You also need to look at the elasticity of supply: in this example, is it easy to enter the search market? You can learn to make a SERP or an algorithm for indexing / ranking results in an algorithms 101 class...but let's not pretend there's not a SIGNIFICANT investment needed in indexing BILLIONS of pages of documents just to be a viable competitor. This is not free by any means, and requires $millions (dozens? hundreds?) of upfront capital just to maybe get a few % points market share.
This is not as convincing for your argument as you may think. All it sounds like is "you took a big risk, you succeeded, and now that we see it's a pretty neat thing, we're going to take it away from you, thanks, please take more risks and make more cool things so we can take them from you".
The fact that this doesn't happen often is why we can have nice things. As soon as you start seizing things that become successful, the nice things stop.
While I _really_ want to agree with you here, in all due respect this is a naive view on antitrust theory.
Elasticity of demand - the ease at which you as a consumer can switch to competitive offerings if price were to suddenly skyrocket - is only one factor of the calculus regarding whether or not Google is anti-competitive.
You also need to look at the elasticity of supply: in this example, is it easy to enter the search market? You can learn to make a SERP or an algorithm for indexing / ranking results in an algorithms 101 class...but let's not pretend there's not a SIGNIFICANT investment needed in indexing BILLIONS of pages of documents just to be a viable competitor. This is not free by any means, and requires $millions (dozens? hundreds?) of upfront capital just to maybe get a few % points market share.
In historical experience, part of the disconnect with the latter piece when discussing this with other tech people is that the easy answer is "oh you can compete with Google: you just need a better algorithm." Of which I'm not sure I can fully agree / disagree - but I _would_ press on the fact that the machine learning scale required would still be a significant upfront investment that makes market entry have steep odds.