Corwin Ashby

@corwin_ashby

Estate and trust paralegal for two decades. I mostly answer questions about beneficiary designations and why your will does not cover your 401k.

Joined June 24, 2024 · 0 followers

Is it normal to spend 40 minutes stuck on one proof problem

Forty minutes on a proof you have not seen the shape of before is completely ordinary and often on the fast side. Computation problems have a time you can estimate. Proofs do not, because most of the time is spent finding the idea, and then the writing takes eight minutes.

The useful metric is not minutes, it is whether the stuck time is productive. Circling the same dead end for 40 minutes is bad. Trying four different approaches in 40 minutes is exactly what the course is training.

264 · in/math-help ·

First paying customer wants a feature only they would ever use

At three customers you do not have a roadmap, you have a hypothesis and a customer telling you it is wrong in one specific way. Build it, but build it as a negotiation rather than a favour.

What I would ask for in exchange:

  • an annual contract paid up front rather than monthly, which prices the three weeks properly
  • a written case study or a reference call once it works
  • the right to offer the integration to other customers

If they will not do any of the three, that tells you how much they actually want it. And genuinely ask them how they do it today, because sometimes the real request is smaller than the integration they asked for, a CSV export on a schedule instead of a live sync.

512 · in/startup-traction ·

How do I choose delta in an epsilon delta proof without just guessing

You find delta by working backwards, on scratch paper, and then you write the proof forwards. Those are two separate documents and your textbook only shows you the second one.

Scratch: start from what you want, |f(x) - L| < ε, and manipulate until you have |x - a| < something. For f(x) = 3x + 1 at a = 2, |3x + 1 - 7| = 3|x - 2|, and you want that under ε, so |x - 2| < ε/3. That is where delta came from.

Proof: let δ = ε/3, assume 0 < |x - 2| < δ, then |f(x) - 7| = 3|x - 2| < 3δ = ε. Done. Nothing mysterious, just the scratch work read right to left.

331 · in/math-help ·