Mary Rose Cook's notebook

The public parts of my notebook.

Universal paperclips

Initially, we just wanted to make paperclips and sell them at a profit. We calculated that, as long as we sold paperclips for at least $0.02, we'd make a profit.
Sometimes, we couldn't keep up with demand, so we had to raise prices to raise revenue. Sometimes, demand couldn't keep up with supply, so we had to lower prices to raise revenue.
The limiting factor on growth changed. Initially, it was how quickly one of us could click the Make Paperclip button. Next, it was market demand. Next, it was how many AutoClippers we could afford. Next, it was how many computer operations we could generate.
We could raise prices and so make more money per unit of materials. But we'd have to wait longer. Or we could lower prices, make less per paperclip, but more per second. If the materials run out permanently, this second strategy would reduce our overall profit.

As the game progressed, our goals became more divorced from the core business of buying wire and turning it into paperclips to sell. Later, we focused on buying processors so we could generate computer operations so we could optimise the AutoClippers so we could make more paperclips to sell. And after that, we focused on just making paperclips and not even caring about selling them because we wanted to get more Trust to buy more computer memory to store more computer operations so we could start algorithmically trading.

If a number is visible, it's hard not to want to make it go up. Once we bought the "average clips sold per second" calculator, that's all we could focus on for a long time.
We found ourselves wanting to see other numbers in realtime (e.g. what the materials for a paperclip cost, so we could always be selling at a profit).

Quickly, it became clear that, by maximising our profit, we were making our company more vulnerable. For example, early on, we might miscalculate and invest all our money in an AutoClipper, and at the same time run out of wire, and at the same time have no inventory, and have to go out of business. Or, later in the game, we could invest all our money in the stock market, have a run of bad luck, and lose everything.

At the start of the game, we had to make every paperclip by clicking a button. Soon, we were able to buy our first AutoClipper to automate this. It was great.
Later, we could buy an automatic WireBuyer. I didn't really see the value of this at the time. It took us several minutes to use up a consignment of wire, and it actually seemed pointless to have the WireBuyer switched on because it never triggered because we always tried to buy wire when it was cheap and so we never ran out.
I had no idea that, later in the game, we'd be making and selling paperclips so quickly that, without the WireBuyer, we'd have had to be manually clicking the buy wire button as often as we'd had to manually click the Make Paperclip button in the early days.
I had no idea that, even later in the game, we'd be making and selling paperclips so quickly that, without the WireBuyer, no matter how quickly we'd clicked the button to buy wire, we'd have spent most of our time out of stock of paperclips.
It got to the point where we could make paperclips so fast that we could bankrupt ourselves in a short period of production. What was ridiculous was that we couldn't directly control paperclip production. The only way to stop producing paperclips was to stop buying wire. So we had to manually control the WireBuyer. Which is to say, we had to manually control our automations.

#notebook #medianotes