How would you identify all the potholes in the United States?

Partner with car manufacturers to get the data from the shock absorbers to directly upload it on to the map. The absorbion can be different based on the kind of car so the complexity will have to be tuned according to the car. For this we need to understand the car suspension system and the manufacturers will need a way to tune the metrics
Self-driving google cars - using its self driving cars it can either take photos and then run through a visualization engine to classify potholes
User generated manual like Waze - Users can input when they go over a pothole. This can be risky if they are driving. Having a voice activated command to the map or waze app to remember the area. Might get missed as people will forget to activate. For this also having a system of feedback from the car to the map automatically.
County offices where the transport records are digitized

Prioritization:

1. COmplexity: Time to get into partnership, accurate way less complexity once metrics have been identified

2. Self driving - low complexity as some part is already being done

3. Waze - Putting voice activated is low complexity, have to be automated either with another device - medium complexity as quality/integration aspects can be tricky.

Things to consider:

Lag between identification and the pothole being fixed.

Updating of the potholes after being fixed will require driving again or updating by the driver or another organization or via self driving cars

1 like | 1 feedback

1 Feedback

Eli Marcus

Platinum PM

Nov 30, 2020

@anand mani I think my biggest problem with your answer is that you neither complete a product design nor do you complete an estimation. If it’s a product design question, then you need to go through the framework of

1. Clarifying questions, When you say potholes do you mean…. I assume that we are only looking at potholes that are on public roads… What is the goal

2. Constraints – I’m only going to think about public roads, etc.

3. Personas

a. Walkers

b. Cyclists

c. Drivers

d. Professional drivers

4. Prioritize the personas

a. I would either focus on drivers or professional drivers

i. This will give us the most data

ii. Professional drivers are always using an app (perhaps google can partner with the apps to gain data).

5. What’s the user journey

a. Gets into car

b. Has phone with them

c. Either goes someplace they know or uses navigation

d. Hits a pothole

e. Spills coffee

f. Curses

6. Pain points (skipping)

7. Priotize pain points

a. This is data collection about potholes so we need to focus on hits a pothole.

8. Brainstorm solutions

a. You came up with a number of quality ideas about how to identify the potholes

i. what about Google maps plus gyro sensors. If I wanted to find all the Potholes in the US I would combine Google Maps plus the phones gyro sensors. I would develop model of what the gryo sensors do when a pothole is struck and then have maps send back a record when it thinks it hits a pot hole I would then compare to other user data and see if we can detect the pattern using machine learning.

9. Compare and contrast

10. Define a metric

11. Summary

If it’s an estimation question

1. Clarify and set boundaries

a. Only public roads

2. Create an equation

a. Potholes per mile * miles of road = potholes in the US

3. Make assumptions

a. Where I live in one mile of road, I encounter 1 pothole

i. Where I live is suburban

ii. Where I live is warm (potholes worse where it freezes)

b. Expand those assumptions to other areas

i. In an urban, warm setting I would expect 2 potholes per mile because…

ii. In an rural, warm setting….

4. Work out the equation and provide an answer – do it on paper without a calculator for practice (I got hung up in my interview not practicing the way the interview was conducted.)

0