3 ms·

Assuming a single data point is representative could add a lot of bias, the cost per trip estimate is probably low given that newer nicer cleaner cars are requi

by hansvm 6y ago

Assuming a single data point is representative could add a lot of bias, the cost per trip estimate is probably low given that newer nicer cleaner cars are required, that assumes drivers are able to drive 100% of the time, it assumes all times of day are equally viable so that there are enough total hours to turn that into a liveable wage, and it's probably a generous representation of tips and a naive view of fees and incentives outside of when a driver first starts. Your numbers might work out, but studies indicate less income in practice, and there are plenty of potential sources of error in your methodology, so I'd be inclined to believe there's something wrong with that ballpark estimate.

AlexTWithBeard 6y ago
> a single data point is representative Please feel free to add your data. I don't think it will change the estimate dramatically. An airport trip takes about an hour and costs $80, which sums up to similar numbers. > that assumes drivers are able to drive 100% of the time No, it doesn't. Reread my message. > it assumes all times of day are equally viable True. I've ignored surge pricing.
- hansvm 6y ago
  > Please feel free to add your data. I don't think it will change the estimate dramatically. An airport trip for me is usually around $15 and 20min. That's in the same ballpark, but even if we get several people from HN to weigh in that doesn't paint a comprehensive picture or account for local variation. > that assumes drivers are able to drive 100% of the time > No, it doesn't. Reread my message. That was my mistake. In my mind I was assuming it would take a car as long to get to me as it would for them to take me where I'm going -- exactly equaling your $30 estimate assuming 0 downtime. Regardless, both the 5min to get to you and 5min of waiting between rides are potential sources of error. Throwing an anecdote into the mix, I've never waited less than 6min for an uber, and I live in a major city. Answering that conclusively needs more data than just the two of us. > it assumes all times of day are equally viable > True. I've ignored surge pricing. And slow periods where drivers wait an hour, late nights where a driver has a higher chance of getting stiffed and needing to clean vomit from their seats, periods just following surge pricing where many drivers have swarmed to an area but the actual available riders have plummeted, and so on. Again, your ballpark numbers might actually be close, but extrapolating a single experience into an entire industry's unit economics is error-prone and IMO irresponsible.

by hansvm 6y ago

AlexTWithBeard 6y ago

> a single data point is representative Please feel free to add your data. I don't think it will change the estimate dramatically. An airport trip takes about an hour and costs $80, which sums up to similar numbers. > that assumes drivers are able to drive 100% of the time No, it doesn't. Reread my message. > it assumes all times of day are equally viable True. I've ignored surge pricing.

hansvm 6y ago
> Please feel free to add your data. I don't think it will change the estimate dramatically. An airport trip for me is usually around $15 and 20min. That's in the same ballpark, but even if we get several people from HN to weigh in that doesn't paint a comprehensive picture or account for local variation. > that assumes drivers are able to drive 100% of the time > No, it doesn't. Reread my message. That was my mistake. In my mind I was assuming it would take a car as long to get to me as it would for them to take me where I'm going -- exactly equaling your $30 estimate assuming 0 downtime. Regardless, both the 5min to get to you and 5min of waiting between rides are potential sources of error. Throwing an anecdote into the mix, I've never waited less than 6min for an uber, and I live in a major city. Answering that conclusively needs more data than just the two of us. > it assumes all times of day are equally viable > True. I've ignored surge pricing. And slow periods where drivers wait an hour, late nights where a driver has a higher chance of getting stiffed and needing to clean vomit from their seats, periods just following surge pricing where many drivers have swarmed to an area but the actual available riders have plummeted, and so on. Again, your ballpark numbers might actually be close, but extrapolating a single experience into an entire industry's unit economics is error-prone and IMO irresponsible.