Examining two assumptions underlying the argument that regulators shouldn’t block consumer tech acquisitions.
What Ben Thompson Assumed
I am a subscriber to, and regular reader of, Ben Thompson’s website, Stratechery. His weekly article this week is entitled ‘First, Do No Harm’. I’m occasionally frustrated by the assumptions that underlie Thompson’s analysis and wish they were more explicit. This article is an attempt to tease those out and scrutinise them just a little.
Unintended consequences only flow from certain actions
Thompson has said at other times that he originally identified as more conservative than he does now. One of the principles from this tradition that he continues to hold is about the risk of actions having unintended consequences. This is commonly deployed as an argument by defenders of the status quo who caution against change that, while well meaning, ends up doing more harm than good. Hence the title of this week’s article.1
One problem with this view is the assumption that only certain actions have unintended consequences. It arises from a quirk in how we speak about ‘action’. Action has consequence and inaction, by implication, does not.2 But since inaction is merely an outcome of the ‘act’ of choosing, inaction is really just a form of action and, well, unintended consequences can flow just as easily from this action as any other. Opponents of action don’t typically use these words but the basis for their position is the belief that the consequences of inaction will be positive. That may not be the case.
A slightly more sophisticated version of the assumption is to take issue with the ‘just as easily’ phrase I used above. Here, one argues that unintended consequences are more likely to flow from actions that seek to change the status quo because we have the control of the status quo against which a comparison may be made. But that’s not true either. We can never run a properly controlled experiment to see what would have happened if the status quo had continued. And without a proper control, we can’t really assess likelihood or magnitude.
The alternative hypothesis is nothing
Issues with hypotheses bring us to the second assumption. Testing hypotheses is at the heart of the scientific method and debaters who wish to clothe their arguments with the authority of science will often use alternative hypotheses as a rhetorical device. I mentioned above that these hypotheses are never actually tested but the second assumption relates to the nature of these hypotheses—or really hypothesis. Because inevitably the alternative hypothesis is that nothing happens.
In the article, Thompson cautions specifically against government intervention in consumer tech acquisitions. Although he is on the record as believing competition regulators should have stopped Facebook’s acquisition Instagram, Thompson argues that this is the exception rather than the rule. Acquisitions are a good thing in consumer tech: they’re good for the general public, they’re good for investors and they’re good for startup founders.
But how can we say whether they’re good for these groups without knowing the alternative? Thompson claims, for example, that they’re good for the general public because the money being poured into consumer tech means that tech reaches the public more quickly than it otherwise would. But this assumes that the alternative is that nothing would have reached those users if not for the service in question. We don’t know that. Why wouldn’t something else have reached them?
If users weren’t using Instagram because it didn’t have Facebook supercharging its expansion, maybe they would have been using a similar service developed in their area. Maybe they would have been using a completely different service superior along some particular vector compared to Instagram. I don’t know. You don’t know. Nobody knows. But we compare the situation to ‘the alternative’ as if we all agree and know what that is.
Similar counters can be deployed to his arguments for investors and startup founders. I’d argue that the implausibility of the ‘nothing happened’ alternative hypothesis is even stronger here. Investors are going to invest in something. Maybe investments in industrial technology would have been better for everyone. Ditto with startup founders who without the exit of a Facebook or Google acquisition as a backstop might have focused on helping people share something other than filtered photos. Somehow technology was developed and disseminated before Silicon Valley tech giants were there to lower the risk that everyone was going to get rich. We might well ask what precisely has this market produced that is considered such an unalloyed good?
None of this is to say that Thompson is completely wrong. You might agree that he’s right in this case. I simply want the assumptions that underlie his argument to be clear. If you agree that unintended consequences can only flow from certain types of action and if you agree that the alternative is that nothing happened, then sure, the argument that we shouldn’t regulate tech acquisitions looks pretty good.
If you don’t accept those assumptions, though? Well, maybe the harm was in not doing anything. ✺
This may seem like a minor point but Thompson’s claim at the beginning of his article that the ‘do no harm’ maxim has the danger of unintended consequences at its core is simply wrong. This is evident from a cursory reading of the Wikipedia article to which Thompson links. I mention it because it’s an example of how the assumptions affect analysis. ↩
This doesn’t actually flow logically but let’s assume for the sake of argument that it does. ↩