Comment from Norway, we have 49 weeks paid family leave full paid up to 70.000$ If the mother is single, you can get up to 3 years of 30.000$ a year. The birth rate is now 1.4, far below the required 2+. As a single mother with many children you can get as money as someone who works here and earns 90.000$ a year (that is without alimony). More money do not solve the birth crisis.
For most women the biggest problem is finding a good man, and they want only the best men (as they have experienced on Tinder) otherwise they will rather have a child on their own than being seen with a man below their standards. For the men, it is really dangerous to get a child with a woman as they compete with the welfare state and have to provide a lot more resources or the mother leaves the man and collect benefits from the state and alimony.
Since there is so much money to be made the women often try to get the children 85-100% to maximise the money they get, and this is pretty easy since you can complain to the police about violence and pedofili and you get full custody. This is pretty normal here but no one talks about it.
As you say this is a lot more complex than more money is good. There is always someone who loses, everything have to be paid by someone else. If you removed all transfers from men to women, the birthrate's would skyrocket, but I do not see that as a viable alternative.
Thank you, great comment. It is indeed very complex to get incentives right, and Norway may be in the "too much" category with the US is in the "too little" category.
Just at a cursory glance, I’m not sure that there’s any correlation between “pro-natalist” or “anti-natalist” government policies and birth rates. I queried ChatGPT and got the following info re: birth rates:
“The 2024 birth rates (per 1,000 people) for the countries you asked about are:
• France: 10.86
• United Kingdom: 11.17
• Germany: 9.32
• Canada: 10.01
• United States: 11.06”
I don’t think that there’s anything to show for paid family leave as being conducive to increased child-bearing. It may be the right thing to do, but I’m not sure arguing that it will boost fertility is the correct argument for making it happen. I don’t see any evidence that those policies boost fertility. I’d rather we argued for it because it’s the right thing to do than argue it based on results that don’t appear in the evidence. Admittedly, this is a cursory check, so there may be evidence that it accomplishes what you’re arguing for, but I’ve never seen any indication of that.
I think it's hard to know because there are multiple variables, and because without these policies these countries may be even lower. We have to think about it in a Bayesian way.
But from a first principle approach, making it more expensive and higher friction to have kids has got to have an impact, just as making it cheaper and easier should, all else equal.
Totally get that. I’m just not sure that there’s not something more visceral going on with birth rates in affluent societies than government policy can really account for. We didn’t need government policies to incentivize what were significantly higher birth rates in the past. I don’t know that there is any government policy that would necessarily move the needle on boosting fertility. Again, that is not to say that we shouldn’t have improved family benefits. This is just a topic I find fascinating.
There are definitely other factors pushing and pulling and they may swamp things like parental leave. But if I'm someone without a lot of money and living paycheck to paycheck and I don't have paid parental leave, it certainly would make me think twice about having kids. May not stop me, but it's one more thing in the way.
I think fertility was much higher before mostly because there wasn't much of a choice about it. Contraception wasn't the same, and women were mostly expected to stay home and raise kids. That started to change when effective contraception became widely available and accepted, and women were given more options.
That's huge chances, and small policie changes won't create huge swings. But as long as we're close or below replacement rate, we should eke out every advantage we can, because they will really matter for humanity longer term. It's like the difference between compounding at 2% or 3% or -2% over a century..
Perplexity is a super valuable tool. While I do not know, it has the appearance of the much maligned "wrapper" type company. In my opinion, this should not be maligned if it is actually true. If it is, they must be losing money hand over fist, praying inference tokens get cheaper. Additionally, they are taking the early Uber approach, build a business and then defend it in the courts at a later date. Finally, based on my calendar, we might be meeting on a Zoom call soon. Looking forward to it.
I ordered a Perplexity t-shirt yesterday, couldn't help it, I really like the company and I hope they succeed and keep building interesting tools.
They are built on top of other people's foundational models, but they seem to be doing enough on top of them to be a bit more than just a wrapper (though I guess it depends how you define what that is, but to my mind, it's more just allowing you to use the vanilla model through some other thing).
They also fine-tune their own models (Sonar) based on Llama, they have their own system to shape results to add citations and make things much more grounded, the whole UI/UX is extremely good and useable, they have a good system to keep track of past searches, they are very quick at getting new models, etc, and afaik, they've been building more and more of their own infrastructure with their recent capital raise, so while at first they were mostly hosted on hyperscalers, they are increasingly a hybrid.
re: Zoom call: I may be missing something, but I don't know what you mean! I'll do another Q&A call with supporters at some point soon, but there's not one planned right now, so unless we have a call related to OSV stuff that I'm not thinking of right now, I'm a bit stumped. Tell me more!
A couple thoughts around the MS/OpenAI relationship; If scaling is simply about adding more/better/faster hardware wouldn't the smart play be to go all in on a single model? Specialized hardware is always going to be more performant for a single use case than generalized hardware.
It could also simply be implying that Microsoft's focus has changed from providing AI services and inference to being an AI cloud provider. If that was the case I'd be less concerned with any specific model and more focused on the generic AI PaaS side of things using what I'd learned from supporting OpenAI.
Maybe that's all just reading too much into what may simply be the clashing of what is likely two very different company cultures and with AI being as hyped as it is, Sam could easily out shadow the MS leadership.
I think when they first partnered, OpenAI felt more ahead of others than it does now. But today, Anthropic, Gemini, Llama, and even Grok feel pretty competitive.
I think Microsoft probably was scared by the whole governance thing and felt like maybe they could lose everything so they decided to hedge their bets now that they saw avenues to do it in a way that is fairly competitive.
But you never know what the future holds. Some AI lab could make some big algorithmic breakthrough and be ahead of everyone else for a while. Which lab? ¯\_(ツ)_/¯
The governance angle is a good point, I could definitely see that playing a role in it as well.
Earlier in the year Scott Galloway and others at ProfG were talking quite a bit about how incestuous the boards seemed to be in this small list of AI and associated companies, and were wondering if the FTC might get involved.
I think that when ChatGPT exploded, Microsoft probably moved VERY quickly to secure a partnership. They probably didn't have a ton of time to consider every angle of it, and hoped that the weird structure of OpenAI wouldn't be a problem... That's my guess
Comment from Norway, we have 49 weeks paid family leave full paid up to 70.000$ If the mother is single, you can get up to 3 years of 30.000$ a year. The birth rate is now 1.4, far below the required 2+. As a single mother with many children you can get as money as someone who works here and earns 90.000$ a year (that is without alimony). More money do not solve the birth crisis.
For most women the biggest problem is finding a good man, and they want only the best men (as they have experienced on Tinder) otherwise they will rather have a child on their own than being seen with a man below their standards. For the men, it is really dangerous to get a child with a woman as they compete with the welfare state and have to provide a lot more resources or the mother leaves the man and collect benefits from the state and alimony.
Since there is so much money to be made the women often try to get the children 85-100% to maximise the money they get, and this is pretty easy since you can complain to the police about violence and pedofili and you get full custody. This is pretty normal here but no one talks about it.
As you say this is a lot more complex than more money is good. There is always someone who loses, everything have to be paid by someone else. If you removed all transfers from men to women, the birthrate's would skyrocket, but I do not see that as a viable alternative.
Thank you, great comment. It is indeed very complex to get incentives right, and Norway may be in the "too much" category with the US is in the "too little" category.
I wrote a bit more about this in a comment here: https://www.libertyrpf.com/p/532-perplexity-buyout-talks-meta/comment/74744345
Just at a cursory glance, I’m not sure that there’s any correlation between “pro-natalist” or “anti-natalist” government policies and birth rates. I queried ChatGPT and got the following info re: birth rates:
“The 2024 birth rates (per 1,000 people) for the countries you asked about are:
• France: 10.86
• United Kingdom: 11.17
• Germany: 9.32
• Canada: 10.01
• United States: 11.06”
I don’t think that there’s anything to show for paid family leave as being conducive to increased child-bearing. It may be the right thing to do, but I’m not sure arguing that it will boost fertility is the correct argument for making it happen. I don’t see any evidence that those policies boost fertility. I’d rather we argued for it because it’s the right thing to do than argue it based on results that don’t appear in the evidence. Admittedly, this is a cursory check, so there may be evidence that it accomplishes what you’re arguing for, but I’ve never seen any indication of that.
I think it's hard to know because there are multiple variables, and because without these policies these countries may be even lower. We have to think about it in a Bayesian way.
But from a first principle approach, making it more expensive and higher friction to have kids has got to have an impact, just as making it cheaper and easier should, all else equal.
Totally get that. I’m just not sure that there’s not something more visceral going on with birth rates in affluent societies than government policy can really account for. We didn’t need government policies to incentivize what were significantly higher birth rates in the past. I don’t know that there is any government policy that would necessarily move the needle on boosting fertility. Again, that is not to say that we shouldn’t have improved family benefits. This is just a topic I find fascinating.
There are definitely other factors pushing and pulling and they may swamp things like parental leave. But if I'm someone without a lot of money and living paycheck to paycheck and I don't have paid parental leave, it certainly would make me think twice about having kids. May not stop me, but it's one more thing in the way.
I think fertility was much higher before mostly because there wasn't much of a choice about it. Contraception wasn't the same, and women were mostly expected to stay home and raise kids. That started to change when effective contraception became widely available and accepted, and women were given more options.
That's huge chances, and small policie changes won't create huge swings. But as long as we're close or below replacement rate, we should eke out every advantage we can, because they will really matter for humanity longer term. It's like the difference between compounding at 2% or 3% or -2% over a century..
Perplexity is a super valuable tool. While I do not know, it has the appearance of the much maligned "wrapper" type company. In my opinion, this should not be maligned if it is actually true. If it is, they must be losing money hand over fist, praying inference tokens get cheaper. Additionally, they are taking the early Uber approach, build a business and then defend it in the courts at a later date. Finally, based on my calendar, we might be meeting on a Zoom call soon. Looking forward to it.
I ordered a Perplexity t-shirt yesterday, couldn't help it, I really like the company and I hope they succeed and keep building interesting tools.
They are built on top of other people's foundational models, but they seem to be doing enough on top of them to be a bit more than just a wrapper (though I guess it depends how you define what that is, but to my mind, it's more just allowing you to use the vanilla model through some other thing).
They also fine-tune their own models (Sonar) based on Llama, they have their own system to shape results to add citations and make things much more grounded, the whole UI/UX is extremely good and useable, they have a good system to keep track of past searches, they are very quick at getting new models, etc, and afaik, they've been building more and more of their own infrastructure with their recent capital raise, so while at first they were mostly hosted on hyperscalers, they are increasingly a hybrid.
re: Zoom call: I may be missing something, but I don't know what you mean! I'll do another Q&A call with supporters at some point soon, but there's not one planned right now, so unless we have a call related to OSV stuff that I'm not thinking of right now, I'm a bit stumped. Tell me more!
ameister@dotadda.com. Email me.
A couple thoughts around the MS/OpenAI relationship; If scaling is simply about adding more/better/faster hardware wouldn't the smart play be to go all in on a single model? Specialized hardware is always going to be more performant for a single use case than generalized hardware.
It could also simply be implying that Microsoft's focus has changed from providing AI services and inference to being an AI cloud provider. If that was the case I'd be less concerned with any specific model and more focused on the generic AI PaaS side of things using what I'd learned from supporting OpenAI.
Maybe that's all just reading too much into what may simply be the clashing of what is likely two very different company cultures and with AI being as hyped as it is, Sam could easily out shadow the MS leadership.
I think when they first partnered, OpenAI felt more ahead of others than it does now. But today, Anthropic, Gemini, Llama, and even Grok feel pretty competitive.
I think Microsoft probably was scared by the whole governance thing and felt like maybe they could lose everything so they decided to hedge their bets now that they saw avenues to do it in a way that is fairly competitive.
But you never know what the future holds. Some AI lab could make some big algorithmic breakthrough and be ahead of everyone else for a while. Which lab? ¯\_(ツ)_/¯
The governance angle is a good point, I could definitely see that playing a role in it as well.
Earlier in the year Scott Galloway and others at ProfG were talking quite a bit about how incestuous the boards seemed to be in this small list of AI and associated companies, and were wondering if the FTC might get involved.
I think that when ChatGPT exploded, Microsoft probably moved VERY quickly to secure a partnership. They probably didn't have a ton of time to consider every angle of it, and hoped that the weird structure of OpenAI wouldn't be a problem... That's my guess
Always interesting....thanks!
Thanks for reading! 💚 🥃