So some people are wondering if whatever the MAGA/Linux crowed is up to is planned and on purpose.
Uncategorized
35
Posts
6
Posters
1
Views
-
@karolherbst @jonkoops that *IS* a couple of orders of magnitude! that's what those bonus multiplications by 10 were for. and this also doesn't include the thermal output from natural radioactive decay in the earth's crust, idk what the order of magnitude there is, but I'm not really expecting it to be small either.
@dotstdy @jonkoops we do grow an OOM of electricity demand each 80 or so years? Accelerated even because we have to replace fossil fuel burning for heating with electricity as well.
I'd have to check up on the real number again, but not sure I'm gonna find it because it's a massive pita to find anything nuclear power related giving how the field is also flooded with bad faith studies.
Maybe it was more like 500 years where it starts to matter... but it's certainly within reach.
-
@dotstdy @jonkoops we do grow an OOM of electricity demand each 80 or so years? Accelerated even because we have to replace fossil fuel burning for heating with electricity as well.
I'd have to check up on the real number again, but not sure I'm gonna find it because it's a massive pita to find anything nuclear power related giving how the field is also flooded with bad faith studies.
Maybe it was more like 500 years where it starts to matter... but it's certainly within reach.
@karolherbst @jonkoops the main point is that the temperature of the earth is an equilibrium between incoming and radiated heat. in such a system the sensitivity to the total incident heat isn't necessarily straightforward. greenhouse gasses change coefficients that affect that equilibrium (as do other things like albedo), and that is why humans can have dramatically outsized impacts on global warming, despite the amount of heat we produce being relatively inconsequential.
-
@karolherbst @jonkoops the main point is that the temperature of the earth is an equilibrium between incoming and radiated heat. in such a system the sensitivity to the total incident heat isn't necessarily straightforward. greenhouse gasses change coefficients that affect that equilibrium (as do other things like albedo), and that is why humans can have dramatically outsized impacts on global warming, despite the amount of heat we produce being relatively inconsequential.
@karolherbst @jonkoops btw as a bonus, the thermal efficiency of a nuclear power station is generally similar to the thermal efficiency of a coal fired power station. so you replace a coal power station with a nuclear one, and you're still rejecting roughly the same amount of heat into the environment. (it's steam turbines all the way down)
-
@ska there is a difference to say that systemd is bad due to irrelevant technical "ideaologies" and having substantial and fundamental criticism about the specific technical choices made.
And I haven't said the criticism is the issue, it's getting rallied into the "anti-systemd" (or whatever" crowd by _agreeing_ on irrelevant points and thinking you make a good point.
If you jump on the "systemd sucks because it's monolithic" hype, you are getting played.
@karolherbst "systemd sucks because it's monolithic" is an oversimplification, but there is some truth to it - it depends on what exactly is meant by "monolithic", a term vague enough that you can attribute a meaning to it and dismiss all the criticism as invalid, when there is a perfectly good argument if we try and define "monolithic" more accurately.
When you think "systemd is obviously not monolithic, these people don't know what they're talking about", you probably mean that systemd is made of several independent amovible parts and not everyone has to run everything. And indeed, under this acception it is silly to accuse systemd of being monolithic.
Here is my fundamental criticism, that I posit is what people mean when they say "monolithic", but that would be more accurately expressed as "hostile" or "closed":
The architecture of systemd, the interfaces it defines, the way it is split into modules, makes it so that it is generally difficult to write a replacement for one of these modules without buying into the systemd model and the way it manages a system. The interfaces are systemd-specific, using concepts that are not used anywhere else than systemd; the modules are designed to interact with systemd, and no effort is made to encourage external software to interact.
As a result, when a replacement module is written, it follows exactly the systemd model, because it's the only, or best, way to make it work. There is no value in writing replacement parts, because they would end up looking exactly like the initial module! The uselessd project was abandoned for this exact reason: it was impossible to make a lightweight clone of systemd.
So, systemd is "monolithic" in the sense that it's a unique piece of software that interacts with itself only; it's not "modular" because no individual module is replaceable, which is one of the points of a modular design.
This post is long enough so I'll stop here, but there are countless examples of hostile design in systemd (I have talked at length about NOTIFY_SOCKET already and am prepared to do more). Because most people are not technical nerds like you and I and are not expressing their criticisms accurately doesn't mean they're entirely wrong in their perception.
I think your callout was using a generic "you", but just in case: I do not jump on any hype, and I am getting played by nobody. I once left a project because it was full of people who were very vocal against systemd without any substance, they were mostly looking to spew hate, and this was not what I wanted from a community. Fortunately, not all alternative communities are like that, and pretending otherwise is doing a disservice to the free software ecosystem.
(Edit: typo)
-
@karolherbst "systemd sucks because it's monolithic" is an oversimplification, but there is some truth to it - it depends on what exactly is meant by "monolithic", a term vague enough that you can attribute a meaning to it and dismiss all the criticism as invalid, when there is a perfectly good argument if we try and define "monolithic" more accurately.
When you think "systemd is obviously not monolithic, these people don't know what they're talking about", you probably mean that systemd is made of several independent amovible parts and not everyone has to run everything. And indeed, under this acception it is silly to accuse systemd of being monolithic.
Here is my fundamental criticism, that I posit is what people mean when they say "monolithic", but that would be more accurately expressed as "hostile" or "closed":
The architecture of systemd, the interfaces it defines, the way it is split into modules, makes it so that it is generally difficult to write a replacement for one of these modules without buying into the systemd model and the way it manages a system. The interfaces are systemd-specific, using concepts that are not used anywhere else than systemd; the modules are designed to interact with systemd, and no effort is made to encourage external software to interact.
As a result, when a replacement module is written, it follows exactly the systemd model, because it's the only, or best, way to make it work. There is no value in writing replacement parts, because they would end up looking exactly like the initial module! The uselessd project was abandoned for this exact reason: it was impossible to make a lightweight clone of systemd.
So, systemd is "monolithic" in the sense that it's a unique piece of software that interacts with itself only; it's not "modular" because no individual module is replaceable, which is one of the points of a modular design.
This post is long enough so I'll stop here, but there are countless examples of hostile design in systemd (I have talked at length about NOTIFY_SOCKET already and am prepared to do more). Because most people are not technical nerds like you and I and are not expressing their criticisms accurately doesn't mean they're entirely wrong in their perception.
I think your callout was using a generic "you", but just in case: I do not jump on any hype, and I am getting played by nobody. I once left a project because it was full of people who were very vocal against systemd without any substance, they were mostly looking to spew hate, and this was not what I wanted from a community. Fortunately, not all alternative communities are like that, and pretending otherwise is doing a disservice to the free software ecosystem.
(Edit: typo)
@ska the general issue here is, that those concepts always have a place. But like you also assume that systemd being more modular would be an advantage by saying that module is better just because.
At which point it's not a great argument to have anymore and ceases to mean anything. Why do we need a more modular design at this point? Systemd can be this tightly integrated project and that's totally fine, even if it's not modular.
It's great we have other init systems tho.
-
@ska the general issue here is, that those concepts always have a place. But like you also assume that systemd being more modular would be an advantage by saying that module is better just because.
At which point it's not a great argument to have anymore and ceases to mean anything. Why do we need a more modular design at this point? Systemd can be this tightly integrated project and that's totally fine, even if it's not modular.
It's great we have other init systems tho.
@ska Like software isn't automatically better just because you can easily replace parts of it with something new following different design principles.
But the point is, those kind of blanket statements do act to pretend having technical knowledge and does give grifters more power than they should have.
I'd ask for more concrete and specific technical criticism, not "this project is bad because it violates the ideas I like personally, but which are academic in nature"
-
@ska Like software isn't automatically better just because you can easily replace parts of it with something new following different design principles.
But the point is, those kind of blanket statements do act to pretend having technical knowledge and does give grifters more power than they should have.
I'd ask for more concrete and specific technical criticism, not "this project is bad because it violates the ideas I like personally, but which are academic in nature"
@ska But there is a fundamental reason why especially alt-right people are so drawn into the anti-systemd crowed: they declare arbitrary "technical" characteristics to declare what software is or isn't superior. "it's not modular, therefore it sucks" does follow fundamentally a supremacy mindset.
They come up wiht a list of what software should be so they can declare their favorite project as the supremacy winner.
That's the dynamic I'm talking about here.
-
@ska the general issue here is, that those concepts always have a place. But like you also assume that systemd being more modular would be an advantage by saying that module is better just because.
At which point it's not a great argument to have anymore and ceases to mean anything. Why do we need a more modular design at this point? Systemd can be this tightly integrated project and that's totally fine, even if it's not modular.
It's great we have other init systems tho.
@karolherbst If you don't see modularity as a goal, then it's fine, but it must also come with the expectation and acceptance that systemd cannot be universal. It should not try to target embedded devices. It should not try to target containers. etc. It cannot have it both ways.
But yes, we probably have a fundamental difference of opinion here: I do believe that one integrated piece of software managing all the parts of a given system is bad design. I believe that several projects interoperating with simple, well-defined interfaces, each one being replaceable, is the superior design, because it gives more power to the user as opposed to the author of the software.
-
@karolherbst If you don't see modularity as a goal, then it's fine, but it must also come with the expectation and acceptance that systemd cannot be universal. It should not try to target embedded devices. It should not try to target containers. etc. It cannot have it both ways.
But yes, we probably have a fundamental difference of opinion here: I do believe that one integrated piece of software managing all the parts of a given system is bad design. I believe that several projects interoperating with simple, well-defined interfaces, each one being replaceable, is the superior design, because it gives more power to the user as opposed to the author of the software.
@ska yeah and my point is that this kind of thinking, if one isn't careful, will mean you might get drawn into alt-right aligned communities, because it's kinda a trope there to declare something as superior based on artificial characteristics.
Does saying "systemd sucks because it's not modular" mean you are part of the MAGA crowd? Absolutely not.
But it's the kind of thinking that aligns with them too well and that's why we do see happening what we see there.
-
@ska But there is a fundamental reason why especially alt-right people are so drawn into the anti-systemd crowed: they declare arbitrary "technical" characteristics to declare what software is or isn't superior. "it's not modular, therefore it sucks" does follow fundamentally a supremacy mindset.
They come up wiht a list of what software should be so they can declare their favorite project as the supremacy winner.
That's the dynamic I'm talking about here.
@karolherbst To be very blunt, I don't give a flying fuck what alt-right people are drawn to. I'm not using that metric as any kind of indicator for software quality, and I don't think anyone should. And if you insist, my personal interpretation of their motives, for most of them, is not "follow a supremacy mindset", it's "be an edgy contrarian and trigger the libs", and for now that means opposing systemd. If s6 was the most deployed init system worldwide, they would probably oppose s6 instead. It doesn't mean anything, I don't care about these people, and you shouldn't either.
-
@ska yeah and my point is that this kind of thinking, if one isn't careful, will mean you might get drawn into alt-right aligned communities, because it's kinda a trope there to declare something as superior based on artificial characteristics.
Does saying "systemd sucks because it's not modular" mean you are part of the MAGA crowd? Absolutely not.
But it's the kind of thinking that aligns with them too well and that's why we do see happening what we see there.
@karolherbst The point that I'm trying to convey here is that the characteristics I'm talking about are not artificial. They have real consequences on how software evolves, from a technical standpoint and also a political standpoint: who wields the power?
I want the power to be in the hands of the user, as much as possible, that is the point of FOSS. And so, I am going to take the position opposite yours: being anti-systemd is a leftist thing to do, because it aims to decentralize power.
The fact that some alt-right people espouse the same position does not mean the position is bad. It means that they feel disempowered too and want more agency. (The difference is that fascists want power over someone less privileged than they are, whereas leftists want justice for all; but that is irrelevant here.)
-
@karolherbst The point that I'm trying to convey here is that the characteristics I'm talking about are not artificial. They have real consequences on how software evolves, from a technical standpoint and also a political standpoint: who wields the power?
I want the power to be in the hands of the user, as much as possible, that is the point of FOSS. And so, I am going to take the position opposite yours: being anti-systemd is a leftist thing to do, because it aims to decentralize power.
The fact that some alt-right people espouse the same position does not mean the position is bad. It means that they feel disempowered too and want more agency. (The difference is that fascists want power over someone less privileged than they are, whereas leftists want justice for all; but that is irrelevant here.)
@ska most users won't feel any difference whether the systems they use run systemd or bash scripts as a system manager.
It's the wrong place to fight this fight.
-
@ska the general issue here is, that those concepts always have a place. But like you also assume that systemd being more modular would be an advantage by saying that module is better just because.
At which point it's not a great argument to have anymore and ceases to mean anything. Why do we need a more modular design at this point? Systemd can be this tightly integrated project and that's totally fine, even if it's not modular.
It's great we have other init systems tho.
Why do we need a more modular design at this point?
because it turns out software developers make mistakes when defining apis and behaviours, and having a way to replace a bad component with a better one is always preferable.
could systemd replace some of its components with some others within itself ? yes. but being modular makes experimentation into potentially better models much easier, and therefore, improvement of the general system also easier.
-
@ska most users won't feel any difference whether the systems they use run systemd or bash scripts as a system manager.
It's the wrong place to fight this fight.
@karolherbst I think you are putting a lot of effort into equating "fighting Nazis" with "being pro-systemd", which conveniently is the system you have chosen to use. Your arguments about "supremacy in software" feel sophistic to me, and could easily be reversed: what piece of software, if any, could be described as hegemonic in the Linux userspace today? 🤔
I don't think we'll come to an agreement over this, so I will let you fight Nazis your way, and I will fight them my way, and I'll be the one choosing the place for that. 🙏
-
undefined oblomov@sociale.network shared this topic
-
@karolherbst @dotstdy @jonkoops BTW if you want some actual numbers I wrote them up some time ago
https://wok.oblomov.eu/tecnologia/nuclear-will-not-save-us/
https://wok.oblomov.eu/tecnologia/nuclear-will-not-save-us-2/
https://wok.oblomov.eu/tecnologia/nuclear-will-not-save-us-3/
Feed RSS
Gli ultimi otto messaggi ricevuti dalla Federazione
-
Va zapp. Che é meglio
-
@rapitadaglialieni la merenda è con i fiocchi ma la cucina è stupenda
-
esattamente! Uguale.
Solo che la serie di cui parlo io era del 1982!
-
Una merenda coi fiocchi!
#SilentSunday
-
io vedevo una serie che si chiamava "Mai dire sì", l'ho ritrovata di recente e la sto rivedendo perchè è di una modernità sorprendente.
Lei una detective che non viene presa sul serio perchè donna, allora si inventa un capo maschio e allora piovono ingaggi. Finchè un ladro (bonissimo Pierce Brosnan) finge di essere questa persona inventata e parte la collaborazione.
Apertura mentale, sociale e moooooolto moderna! 😜
-
@DagwoodIII In my opinion, the point is even more serious:
1) Company value increases to the extent that financiers (stockholders, investment funds, crowdsourcing) believe the sector can multiply their investment.
2) All companies (not just AI companies) know that their value increases with the number of layoffs (lower fixed costs, higher margins).
3) Particularly in the AI sector, company value increases with the purchase of new Nvidia hardware and the opening of new data centers.
Therefore, the combination of 2) and 3) leads to a physiological depression in AI engineering hiring: it's better to hire fewer systems engineers and pay the companies that supply software and build data centers.
But these are logics that conflict with the real development of that sector!
In Italiano:
Il punto secondo me è ancora più grave:
1) il valore delle aziende aumenta nella misura in cui i finanziatori (azionisti di borsa, fondi di investimento, crowdsourcing) credono che il settore possa moltiplicare il loro investimento
2) tutte le aziende (non solo qulle di intelligenza artificiale) sanno che il proprio valore aumenta con il numero di licenziamenti (meno costi fissi, più margini)
3) in particolare nel settore dell'Intelligenza Artificiale, il valore delle aziende aumenta a fronte dell'acquisto di nuovo hardware Nvidia e nell'apertura di nuovi datacenterpertanto, il combinato di 2) e 3) comporta una fisiologica depressione delle assunzioni di ingegneri AI: meglio assumere pochi sistemisti e pagare le ditte che riforniscono software e costruiscono datacenter.
Ma queste sono logiche che vanno in conflitto con uno sviluppo reale d quel settore!
-
Stessa lista, ma io pure Mork&Mindi con un giovanissimo Robin Williams. Mimica spettacolare!
-
Oddio che ricordonche mi avete sbloccato!!!
