Expected Steps in Uniform Process

This quant interview question is about a random waiting time in a simple stochastic process where the success chance itself is randomly drawn. It links a basic geometric-type setup with an extra random layer, turning an apparently easy expected value into something more subtle. Candidates see how a simple uniform choice can create a nontrivial distribution for the stopping time.

It trains conditional expectation, the Law of Total Expectation, and comfort with random parameters inside probability models. It also sharpens intuition about how expectations behave when you integrate over a distribution that has significant mass near dangerous regions. That is central to serious quant prep, beyond textbook plug-and-chug.

This matters for quant interviews because it probes whether you recognize when expectations can blow up, how you think about random success probabilities, and whether you can connect elegant theory to quick mental diagnostics in real interviews.

What it tests

When a random process's success probability is itself a random variable, the overall expectation is found by averaging the conditional expectations over the distribution of that parameter. This is a direct application of the Law of Total Expectation: $\mathbb{E}[N] = \mathbb{E}[\mathbb{E}[N \mid P]]$, where `P` is the random success probability. The key is that the 'average of reciprocals' (e.g., $\mathbb{E}[1/P]$) can behave very differently from the 'reciprocal of the average' (e.g., $1/\mathbb{E}[P]$), especially when the distribution of `P` puts weight near zero, causing divergence. This principle explains why conditioning on a random parameter and then integrating can reveal infinite expectations, even when each conditional expectation is finite. The divergence often arises because the random parameter can get arbitrarily close to a value (here, zero probability of success) that makes the expected waiting time blow up.