That promotion was never going to work if people could open an endless number of accounts, but how do you prevent people from creating multiple accounts on a free platform to benefit from a joiner's discount?
This is one of the major reasons why free subscriptions are such a bad idea. You could end up with a disproportionate amount of users using the free version because of duplicate accounts. One way to avoid this is to charge, but beyond that, if you are going to offer multiple subscription options, make sure your paying options don't just offer more of the same - make sure they offer different features and/or services - something a user can't access no matter how many free accounts it opens.
I think this is fine for testing and idea validation as long as the long term intention is to move it to an AI. I'm sure hiring workers like that isn't sustainable in the end.
If you showcase with the "Wizard of Oz" strategy that there's interest and can then indicate progress and ability to deliver on the AI I think it's fine
I really like the "behind the curtain" approach as its the absolution of "do things that dont scale". That was one of the most valuable things i learned from the essays of Paul Graham.
I've heard this pattern described as "Rooms full of people". Usually people say they are doing RFOP as a lead up to the actual model that does the work, but if you don't communicate that... lol
All jokes aside, while he lied, maybe he's taking the test-driven approach of getting expected answers (from people) and writing an algorithm to emulate people.
another thing is maybe something was working with AI, but corner cases needed people. I read somewhere Alexa is was still validated by people sometimes...
It reminds me of the blood-testing startup which got busted at the start of this year.
That promotion was never going to work if people could open an endless number of accounts, but how do you prevent people from creating multiple accounts on a free platform to benefit from a joiner's discount?
This is one of the major reasons why free subscriptions are such a bad idea. You could end up with a disproportionate amount of users using the free version because of duplicate accounts. One way to avoid this is to charge, but beyond that, if you are going to offer multiple subscription options, make sure your paying options don't just offer more of the same - make sure they offer different features and/or services - something a user can't access no matter how many free accounts it opens.
Intentional deception for gain is called fraud.
Wizard of Oz as a tactic is fine, but lying about how your company works is not.
I think this is fine for testing and idea validation as long as the long term intention is to move it to an AI. I'm sure hiring workers like that isn't sustainable in the end.
If you showcase with the "Wizard of Oz" strategy that there's interest and can then indicate progress and ability to deliver on the AI I think it's fine
I really like the "behind the curtain" approach as its the absolution of "do things that dont scale". That was one of the most valuable things i learned from the essays of Paul Graham.
This is common. See the "Pretotyping" book, many startups pretended to have software do the job when in fact it was people in the background.
I've heard this pattern described as "Rooms full of people". Usually people say they are doing RFOP as a lead up to the actual model that does the work, but if you don't communicate that... lol
AI is just bunch of if/else though...
All jokes aside, while he lied, maybe he's taking the test-driven approach of getting expected answers (from people) and writing an algorithm to emulate people.
another thing is maybe something was working with AI, but corner cases needed people. I read somewhere Alexa is was still validated by people sometimes...
reminds me of this tweet: https://twitter.com/naval/status/1225674500882100226
How are they supporting this volume of sign-ups with people?! I don't understand how it's even possible https://startupsavant.com/news/nate-fintech
This comment was deleted a year ago.