I agree with your comments when scoped to private tech/saas. The tech recurring revenue model has a lot of volatility compared to other recurring revenue models in other industries.

Your statement is better than the OP but still not correct. A private company that is not profitable can STILL have a valuation. You are forgetting that it can have assets like IP, personnel, real estate, equipment etc. An example is a licensed insurance shell company with an operating loss, that can sell for millions due to the value of the existing licenses + in place admin. The present value of the future value of the company based on relation to it's peers can sometimes also factor in, especially if there is a small capital need to bring it up to a profitable level, considering the business model is sound.

Shockingly enough, there are a lot of very intelligent people who excel in their fields but know nothing about how these ML models are put together. To them it is something that is plausibly true. A personal example is a cousin of mine who is a technical finance guy that brought this up. I showed him two articles that explained why it is not sentient or artificial general intelligence. He was open to learning and then understood that while it's a great technical achievement, it's far from what this guy was claiming.

How is 2.x a better language? Any specific examples? That is a pretty broad generalization and not really supported by a lot of the python community, except maybe the very vocal minority. I can see some things in 3.x that were done to be more consistent, example would be the print statement that uses () instead of space. Yes, it did break backward compatibility. I guess the question would be how much of a problem did this change actually make for you? A lot of people complain about things that in reality had no/minimal impact to them. Was there something that caused headaches for your migration, or are you just talking about backwards compatibility in general?

I agree that backwards compatibility is important, but doesn't have to be in every case. Also, most real work is been done on python 3.x not 2.x. If you are still using 2.x or doing new development on it then you are rolling the dice on security. It's 2019, there is really no reason to use 2.x, unless you lack the resources to migrate your production app or have an edge case with a library, which means you probably should have chosen Java or .net in the first place.

This seems to be a big enough decision for a ballot measure. Even though it would be nice to have that factory here, I think it is a mistake to make all those concessions.