It is easy to delude yourself in abstractness.

To take a personal example: I spent three months building Anysphere, a metadata-private communication platform. My personal motivation for doing it was that I imagined that creating metadata-private communication would increase the probability that certain authoritarian regimes are overthrown. The business motivation was that companies and democratic governments care about security, and hence they should be willing to spend money on the most secure communication tool possible. Convinced? I was, because I did not yet see the danger of abstract thinking.

Neither my personal nor the business motivation holds up to concrete scrutiny. If our platform ever came close to threatening the Chinese government, it would instantly be blocked. Thinking concretely about what problems activists in China actually face, rather than abstractly saying that obviously it would be useful, would have helped me see that metadata privacy alone achieves very little without censorship resistance. Similarly, if we had been thinking concretely about exactly how our high-latency communication platform could be used to help organizations be more secure, we would have realized that our platform didn’t solve any actual problems that organizations have (except in certain particular cases, e.g. for spies), and might in fact worsen security by making human error easier.

If you’re a person with too much pride (as I am), the delusional effect of not being concrete gets even worse. People will sense your pride, and refrain from asking questions that force you to make your vision concrete. There’s also often a touch of it’s obvious in abstract language, especially if it’s coming from a person who is seen as knowledgeable in the relevant domain. This makes clarifying questions intimidating to ask. In hindsight, regretfully, I feel that many of my conversations about Anysphere fell victim to this.

Now, what do I propose? Let’s be concrete about being concrete. In the future, whenever I have a plan or an idea, I will spell out all the details. No hand-waving. “Anysphere will be used to send sensitive documents from whistleblowers who want to remain secret to journalists, after establishing contact over email” rather than “Anysphere will be used by journalists”. “Access to Anysphere will be censored if it would ever be useful as a democracy-inducing tool” rather than “Anysphere will create democracies”.

In markets, concreteness is naturally favored. When no one was willing to pay for metadata privacy, we were forced to come out of our bubble of abstractness. This is good. In other domains, such as academia, there is no immediate selection for concreteness. Many papers in the field of metadata privacy make sweeping claims about how bad surveillance is in the introduction, making it hard to realize that the solutions they propose will never solve that problem in practice. While it certainly can be worth doing research for the sake of exploring, one should be fully aware of the limits of such research, remembering that it is unlikely for abstract thinking to lead to concrete results.

This has been said before, better, by George Orwell, concerning the abstract uses of language. Slipping away from concreteness is easy, but harmful. Personally, I will try hard to stay concrete.