4 ms·

> The PSF is a relatively small organization, operating with an annual budget of around $5 million per year, with a staff of just 14. This might be the bigger

> The PSF is a relatively small organization, operating with an annual budget of around $5 million per year, with a staff of just 14. This might be the bigger story. How many trillions of dollars depend on Python? Yes, I mean trillion. Those market caps didn't skyrocket on nothing. A lot of ML systems run on Python. A lot of ML systems are first implemented in Python. Even with more complicated backends a Python layer is usually available, and used. A whole lot of other stuff depends on Python too, but the AI part is obvious. This is the weird part about our (global![0]) economics that I just don't get. We'll run billions of dollars in the red for a decade or more to get a startup going yet we can't give a million to these backbones? Just because they're open source? It's insane! If we looked at projects like this as a company we'd call their product extremely successful and they'd be able to charge out the wazoo for it. So the main difference is what? That it's open source? That by being open source it doesn't deserve money? I think this is a flaw we probably need to fix. In the very least I want those devs paid enough that they don't get enticed by some large government entity trying to sneak in backdoors or bugs. [0] it's not just the US, nor is it just capitalist countries. You can point me at grants but let's get honest, $5m is crazy low for their importance. They're providing more than 1000x that value in return. [side note] I do know big companies often contribute and will put a handful of people on payroll to develop, bug hunt, etc. But even if we include that I'm pretty sure the point still stands. I'm open to being wrong though, I don't know the actual numbers [P.S.S] seems to parallel our willingness to fund science. Similarly people will cry "but what is the value" from a smartphone communicating over the Internet, with the monetary value practically hitting them in the face.

braza 1y ago
> A lot of ML systems run on Python. A lot of ML systems are first implemented in Python. > That by being open source it doesn't deserve money? I think this is a flaw we probably need to fix. Independent of how one feels about the current US administration, I do not think, as a non-American, that a particular government should foot the bill for it, but in reality I know that no company will do it in good will either. I've been thinking a lot in terms of financing, but the current system of grants, where some agency tied with the executive body will approve or reject something, is fundamentally broken, as we can see. In those cases of critical infrastructure, I think it's worth some kind of minimum 1:1 deductible of pre-tax programs where the foundations can apply, and then they could have their financing without being at the whims of some branch of the executive.
- godelski 1y ago
  > I do not think, as a non-American, that a particular government should foot the bill for it It is definitely a complicated problem but governments tend to but good funding agencies for work that uplifts the broader society and creates the foundation for new markets. That's the idea behind science funding anyways. New science might not create a trillion dollar business directly but it sure lays the funding for new multi billion dollar companies and companies to skyrocket from 500bn to 5T market caps... But my point is that a project like this is global. I want the US putting money in. We're the richest and benefiting the most. But I also want other countries putting money in. They should have a vested interest too. I think an interesting mechanism might be to use agencies like the NSA. We know their red teams but what about the blue? I'd love for the blue teams to get more funding and have a goal to find and patch exploits, rather than capitalize on them. Obviously should have a firewall between the teams. But this should be true for any country. It might just be some starting point as it could be a better argument for the people that don't already understand the extreme importance of these types of open source projects. > I think it's worth some kind of minimum 1:1 deductible of pre-tax programs Typically these projects run as nonprofit foundations. They're already getting tax benefits. Though I think we can recognize that this isn't enough and isn't remotely approaching the value. It's definitely not an easy problem. Like what do you do? Tax big companies (idk, an extra 0.1%?), audit to determine dependencies, distribute those taxes accordingly? In theory this should be simple and could even be automated, but I'm sure in the cat and mouse game the complexity would increase incredibly fast. But hey, it shouldn't just be America. Different countries can try different ideas
nomdep 1y ago
The PYthon Software Foundation does not develop Python. They manage the package repository, do marketing pro-Python, and distribute money among people they like (in theory to promote the language). 10 of the 13 people of their staff doesn’t even know how to write a hello world program https://www.python.org/psf/records/staff/ https://www.python.org/psf/records/staff/
BrenBarn 1y ago
Yes, it's crazy. I think a lot of people see it as a question of "how can we give the PSF (or orgs like it) more money" but I see it a bit differently. Basically if something like Python can arise and become so effective and useful in so many ways with so little funding (and even less in earlier stages), it suggests that money isn't really the bottleneck here. What we need are people doing good work with good motives and not chasing dollars. That in turn suggests that a lot of money currently being spent is wasted, or worse, used for ill. We would be better served by taking all the assets of the Fortune 500 and distributing them widely to tons of little groups. Some of those groups may turn out to be the next Python, and for the ones that don't, well, we didn't waste much money on them. Right now what we get instead is hundreds of billions of dollars going to advertising algorithms. The reason it's crazy that the PSF survives on $5 million isn't that $5 million is crazy little, it's that too many other entities are crazy big.
- godelski 1y ago
  That is another great point. I do see a lot of companies trying to cut corners and make things as cheap as possible but I fundamentally believe that this ends up increasing costs in the long run. Which is a bit weird since I know a lot of business people and most seem to be familiar with Boot Theory[0] and even agree with it. But maybe it is one of those things that is just not realized in practice. Yet I constantly see these sacrifices being made. Profits over product. I don't think that actually leads to more profits. Maybe in the quarter, but not long term. It's a sustainable strategy if you've monopolized the market though... I also notice that when layoffs come around that the people who decide who to lay off tend to not lay off themselves. Great way to become management heavy... When in reality, if you have a lot of "rockstar programmers" they should be able to mostly manage themselves. You want to keep them from going down too many rabbit holes and help them prioritize, but for the most part experts know what is most important in the system they're working in. [0] https://en.wikipedia.org/wiki/Boots_theory https://en.wikipedia.org/wiki/Boots_theory
  - BrenBarn 1y ago
    > Maybe in the quarter, but not long term. It's a sustainable strategy if you've monopolized the market though... It's also a sustainable strategy if the market has become so distorted that it's difficult for consumers to distinguish the good products from the bad. This is a pervasive problem these days. It may still be the case that a good pair of boots will last you 10 years, but how can you tell which pair is the good one? There are 10,000 brands, all of them have 4.7 star ratings because of bots, most of them have a picture that looks nothing like what you'll receive, and if they wear out in two years you'll have no recourse because the brand was just a facade over a "ghost workshop" that's now producing the same crap under another name. Trying to vote with your dollars is like pushing on a rope. > When in reality, if you have a lot of "rockstar programmers" they should be able to mostly manage themselves. I'm not so sure about this. There are many people who are good at doing things but not at organizing that activity in a coherent way (especially across a team). Also when the people doing the "real work" are also managing, they often have a tendency to give the managing short shrift because they'd rather do the "real work". I agree we could do with a lot fewer managers than we currently have, but I think we do need some. (And this isn't just about software, it's about companies and organizations in general.) But I think we would be better off with more democratized company structures where the managers' power ultimately rests on the consent of the workers, so that when you have these clashes that currently result in "half the staff resigned in protest" they would instead result in "the CEO was fired due to lack of support from the staff".

braza 1y ago

> A lot of ML systems run on Python. A lot of ML systems are first implemented in Python. > That by being open source it doesn't deserve money? I think this is a flaw we probably need to fix. Independent of how one feels about the current US administration, I do not think, as a non-American, that a particular government should foot the bill for it, but in reality I know that no company will do it in good will either. I've been thinking a lot in terms of financing, but the current system of grants, where some agency tied with the executive body will approve or reject something, is fundamentally broken, as we can see. In those cases of critical infrastructure, I think it's worth some kind of minimum 1:1 deductible of pre-tax programs where the foundations can apply, and then they could have their financing without being at the whims of some branch of the executive.

godelski 1y ago
> I do not think, as a non-American, that a particular government should foot the bill for it It is definitely a complicated problem but governments tend to but good funding agencies for work that uplifts the broader society and creates the foundation for new markets. That's the idea behind science funding anyways. New science might not create a trillion dollar business directly but it sure lays the funding for new multi billion dollar companies and companies to skyrocket from 500bn to 5T market caps... But my point is that a project like this is global. I want the US putting money in. We're the richest and benefiting the most. But I also want other countries putting money in. They should have a vested interest too. I think an interesting mechanism might be to use agencies like the NSA. We know their red teams but what about the blue? I'd love for the blue teams to get more funding and have a goal to find and patch exploits, rather than capitalize on them. Obviously should have a firewall between the teams. But this should be true for any country. It might just be some starting point as it could be a better argument for the people that don't already understand the extreme importance of these types of open source projects. > I think it's worth some kind of minimum 1:1 deductible of pre-tax programs Typically these projects run as nonprofit foundations. They're already getting tax benefits. Though I think we can recognize that this isn't enough and isn't remotely approaching the value. It's definitely not an easy problem. Like what do you do? Tax big companies (idk, an extra 0.1%?), audit to determine dependencies, distribute those taxes accordingly? In theory this should be simple and could even be automated, but I'm sure in the cat and mouse game the complexity would increase incredibly fast. But hey, it shouldn't just be America. Different countries can try different ideas

nomdep 1y ago

The PYthon Software Foundation does not develop Python. They manage the package repository, do marketing pro-Python, and distribute money among people they like (in theory to promote the language). 10 of the 13 people of their staff doesn’t even know how to write a hello world program https://www.python.org/psf/records/staff/ https://www.python.org/psf/records/staff/

BrenBarn 1y ago

Yes, it's crazy. I think a lot of people see it as a question of "how can we give the PSF (or orgs like it) more money" but I see it a bit differently. Basically if something like Python can arise and become so effective and useful in so many ways with so little funding (and even less in earlier stages), it suggests that money isn't really the bottleneck here. What we need are people doing good work with good motives and not chasing dollars. That in turn suggests that a lot of money currently being spent is wasted, or worse, used for ill. We would be better served by taking all the assets of the Fortune 500 and distributing them widely to tons of little groups. Some of those groups may turn out to be the next Python, and for the ones that don't, well, we didn't waste much money on them. Right now what we get instead is hundreds of billions of dollars going to advertising algorithms. The reason it's crazy that the PSF survives on $5 million isn't that $5 million is crazy little, it's that too many other entities are crazy big.

godelski 1y ago
That is another great point. I do see a lot of companies trying to cut corners and make things as cheap as possible but I fundamentally believe that this ends up increasing costs in the long run. Which is a bit weird since I know a lot of business people and most seem to be familiar with Boot Theory[0] and even agree with it. But maybe it is one of those things that is just not realized in practice. Yet I constantly see these sacrifices being made. Profits over product. I don't think that actually leads to more profits. Maybe in the quarter, but not long term. It's a sustainable strategy if you've monopolized the market though... I also notice that when layoffs come around that the people who decide who to lay off tend to not lay off themselves. Great way to become management heavy... When in reality, if you have a lot of "rockstar programmers" they should be able to mostly manage themselves. You want to keep them from going down too many rabbit holes and help them prioritize, but for the most part experts know what is most important in the system they're working in. [0] https://en.wikipedia.org/wiki/Boots_theory https://en.wikipedia.org/wiki/Boots_theory
- BrenBarn 1y ago
  > Maybe in the quarter, but not long term. It's a sustainable strategy if you've monopolized the market though... It's also a sustainable strategy if the market has become so distorted that it's difficult for consumers to distinguish the good products from the bad. This is a pervasive problem these days. It may still be the case that a good pair of boots will last you 10 years, but how can you tell which pair is the good one? There are 10,000 brands, all of them have 4.7 star ratings because of bots, most of them have a picture that looks nothing like what you'll receive, and if they wear out in two years you'll have no recourse because the brand was just a facade over a "ghost workshop" that's now producing the same crap under another name. Trying to vote with your dollars is like pushing on a rope. > When in reality, if you have a lot of "rockstar programmers" they should be able to mostly manage themselves. I'm not so sure about this. There are many people who are good at doing things but not at organizing that activity in a coherent way (especially across a team). Also when the people doing the "real work" are also managing, they often have a tendency to give the managing short shrift because they'd rather do the "real work". I agree we could do with a lot fewer managers than we currently have, but I think we do need some. (And this isn't just about software, it's about companies and organizations in general.) But I think we would be better off with more democratized company structures where the managers' power ultimately rests on the consent of the workers, so that when you have these clashes that currently result in "half the staff resigned in protest" they would instead result in "the CEO was fired due to lack of support from the staff".