Do Things That don't Scale

Paul Graham wrote an essay discussing the importance of doing things that don't scale when you are building a startup (he defines them as companies in a larval state). However, I have seen the sentence "do things that don't scale" applied in very weird ways, possibly by people who never stopped to think what it actually means.

If you go through the original article, you see that most of the focus is payed on how you can acquire users. It cites the example of AirBnb , that started by having the founders going to take photos of the apartments themselves. There is another common example of someone delivering shoes.

However, this is greatly misleading. AirBnb is still taking photos of apartments, and the delivery of clothes didn't have any kind of disruption on the delivery process: someone needs to physically bring the goods from a warehouse to your doorstep. The scaling in this respect, only relates to hiring people to do the work for you.

On the other hand, user acquisition is also discussed, and is the paramount argument of the essay. If you need to spend dozens of hours to convince someone to use your product, it will take forever to have enough users to be able to survive. This is the part doesn't scale . However, once you are known, users will flock to your product with much lower effort from your side.

What I found people not understanding is how to separate what is meant to scale from what is not meant to scale. AirBnb is just a website, no one doubts it can be built and the needed tools are already available. However, such as the case with Uber , another mastodon example, the true unscalability of the business comes from legal regulations.

You may bribe (or call it lobby) legislators in some cities. Are you sure you can really scale that up to the entire world?

A very different approach of scaling or not, is what happens when the infrastructure and the technical tools are not there. Streaming, such as Netflix does, is possible thanks to Amazon 's infrastructure and a global improvement on internet access. If they would have started 20 years ago with a streaming service, hosting only 10 movies on a server, could they have scaled?

My bottom line about doing things that don't scale is that you should do them provided that you know they can scale. If you already know that there is a hard limit, for example that regulations (or just common humanity to your own kin) do not allow you to underpay taxi drivers, would you still be so valuable in the eyes of Paul Graham ?

I see a very big similarity between "doing things that don't scale" and being lean . That is why I think the original essay is quite empty on actionable knowledge and it lacks original insight. It even mentions Facebook because they narrowed down their launching market to just one university. I think just throwing things to the mix to show how cool and well connected you are does not provide value but to your own persona (and by extension your company).

I found it very interesting that it even dares putting into the mix hardware startups, using Pebble as an example. It cites that the founders were making the watches themselves as an example of non-scalability. This is plainly ridiculous. Is the fact that they did it themselves instead of an underpaid chinese employee in a distant factory an example of something that does not scale?

It could have gone deeper on any of the matters at hand. I'm not sure of the details, but with Pebble a better example could have been using 3D printed parts instead of injection molded, or THT electronics instead of SMD.

Users and but not Customers

Something that strikes me and that I've seen very often on Y-Combinator materials is that they almost always talk about users, and not about customers. Facebook needed years to understand that their value, the things they had to sell were their users' data. They talked about AirBnb users and not clients . I think this is a very biased view of the world, that washes away the responsibility they have.

If you are a user of a platform, the platform is neutral . If you do something illegal, such as renting an apartment on a temporary base even if not allowed, the platform only enables, does not regulate. You can sell your users' data, it is only a problem of whoever buys it to make morally correct things.