mpweiher

You might need to adjust your sources.

Do you have any evidence for your claim?

This meme that nuclear power plants are unprofitable and require subsidies has been spread wide and far by the anti-nuclear lobby, but is still false.

As an example, when the Greens started to come into government in Germany, they tried to force nuclear operators out of business by creating onerous safety rules just for the purpose of forcing them out of business.

It didn't work.

The plants were so profitable that they could afford even the most ridiculous additional safety requirements. So they had to make them illegal.

Furthermore, you can look at France. EDF has been immensely profitable, despite having to sell a large amount of electricity at discounted rates.

Here's an explainer.

https://www.youtube.com/watch?v=cbeJIwF1pVY

> Sadly, the ABWR has also proven to be unreliable and uneconomic.

Citation needed.

> Hitachi spent most of the 2010s trying to get a couple of them underway in the UK

That's not evidence of them being uneconomic or unreliable.

> (which has a generally favourable regulatory environment)

Excuse me? The regulatory environment that is responsible for the 7000 design changes at Hinkley Point C and thus most of the eye-watering delays and cost overruns? The regulatory environment that required the £ 700 million "fish disco" that will save a few salmon at a cost of around £ 280000 per fish?

A series of "radical, root-cause solutions" is required to simplify the UK's nuclear regulatory system in order to speed up the construction of new nuclear projects at a lower cost and on time, an independent taskforce has concluded.

https://www.world-nuclear-news.org/articles/radical_reforms_...

Yes. Most nuclear power plants are (highly) profitable and not subsidized.

This is incorrect.

> That's some serious cherry picking you're doing there.

Please identify the cherry picking.

> The LCOE calculations also do not capture other systemic costs or externalities beyond plant-level CO2 emissions such as, for instance, methane leakage during the extraction and transport of natural gas."

Absolutely! When you consider full system costs, nuclear gets much, much better.

With fossils, you have the minor externalities of climate change and other emissions, which are not taken into account.

With intermittent renewables, you have the costs of their intermittency. Those system costs tend to rise with penetration, and dwarf the LCOE.

For details in a model, see:

https://www.sciencedirect.com/science/article/pii/S036054422...

And this also plays out in real life: electricity prices are highest in countries with high intermittent renewables penetration. The correlation is quite strong. And the inverse correlation, from high penetration to low electricity costs simply does not exist. There are no countries (or states) with high penetration of intermittent renewables and low electricity prices.

> So we can just gloss over the nuclear waste problem.

The opposite is true. Nuclear is actually the only power source that has to account for its waste already in the LCOE cost. And it also turns out that the waste is one of the benefits of nuclear, at least compared to other sources of electricity: there is very, very little of it, we know how to store it safely without problems, it actually goes away by itself and it is valuable fuel.

[Costs from 2-3 cents to below 1 cent]

> Startups have cost projections, sure. That's marketing material until they've actually built something.

Nope. Those numbers are usually the sorts of things they have to present investors to make their business case. People tend to vet those numbers pretty carefully before investing millions or billions of dollars. Also, if you lie in those numbers that turns out to be fraud and you can go to jail.

In other news: "Prediction is difficult, especially about the future" -- Yogi Berra.

Nuclear reactors are highly profitable:

https://www.youtube.com/watch?v=cbeJIwF1pVY

Lowest LCOE by far is "nuclear LTO (Long Term Operation)".

https://www.iea.org/reports/projected-costs-of-generating-el...

Yes, the plants are big and expensive, but once built, they are cheap to run and last pretty much forever.

SMRs lower the up-front cost, the time to build, the risk, and the financing costs, which are the biggest component of the construction costs.

Initially at somewhat higher cost per kWh, but there is plenty of headroom there. And the various nuclear startups have cost projections that range from 2-3 cents to below 1 cent / kWh.

Smaller reactors = more reactors.

More reactors = riding the cost curve more quickly.

The label SMR applies to a wide range of reactors.

The BWRX-300 is at the upper end of that range and I don't think claims of factory-production of the whole unit were ever made for this reactor.

That said, even the much larger AP-1000 had fairly large modules made in a factory. In fact as far as I understand that was one of the problems with the Vogtle builds, because doing that only really makes sense for a larger number of units, not for just two unites.

Hmmm...is it just me or are pretty much all of the recent announcements from the AI companies about lower prices for comparable performance?

So we've already reached the commodification phase, due to either the frontier models not getting usefully better or the models already being good enough, or both.

In either case, there appears to be a plateau in performance. With local open-source models catching up.

Interesting times.

How certain are you of your prediction? What odds would you give me if I bet against you?

10:1?

100:1?

Background:

The BWRX predecessor, the ABWR, holds the record for the fastest construction time of a commercial nuclear power plant ever: just slightly over 3 years to first criticality, 4 years total to commercial operation.

Fun fact: it was the success of this first Gen III reactor that caused EDF to predict the EPRs would also only take 3 years to build. Which proved...optimistic. For the EPR. But proven for the ABWR.

https://en.wikipedia.org/wiki/Advanced_boiling_water_reactor

https://en.wikipedia.org/wiki/Kashiwazaki-Kariwa_Nuclear_Pow...

https://hannahritchie.substack.com/p/nuclear-construction-ti...

The BWRX is also passively safe: cooling occurs via natural circulation, no pumps needed.

So if it takes 15 years I give you $100, if it takes less you give me $10000?

Deal?

> So UI matters enough to justify complete rewrites, but also everyone just wants CLIs?

Sloppy.

DHH said that he also wants CLIs. Not "just". Most of the rest is actually very similar.

No he's not talking about race.

You are.

And he's not talking about deporting citizens.

You are.

Yes, and?

Says nothing about (a) race or (b) citizens.

On the other hand, he specifically writes about "hard problems like cultural integration between wildly different ways of living."

So culture, not race.

So exactly as I wrote: he talks about non-citizens and you turn it into "race".

Who is the racist?

1. Whether that is OK or not is unrelated to the question of whether it is racist to revoke citizenship on non-racial grounds. Which it clearly is not. (The "...the wrong race" is purely you).

2. As far as I can see, the position that DHH referred to did not apply to naturalized citizens, but to foreigners. So non-citizens.

3. As far as I know, most countries currently have laws on the books that can result in citizenship being revoked. As an example, countries that don't allow dual citizenship will take your citizenship away if you become citizen of another country. Germany and many other countries will take your citizenship away if you lied on your application and it only comes out later. Terror and treason are also common reasons.

4. Some counties (India for example) will even take your (naturalized) citizenship away if you are out of country long enough.

That's kind of the point of conference talks.

"Here's something that I did that turned out to be helpful to my situation. Maybe you can learn something useful from it for your situation"

Where did he openly support "ethnic cleansing via remigration"?

Googling, I can find this:

https://x.com/dhh/status/1997275678555951451

Which refers to this:

https://www.bt.dk/samfund/brandfarligt-ny-maaling-viser-mass...

Both of these have a different definition of "remigration" than what you are using:

- deporting foreign nationals, not legal citizens

- deporting those foreign nationals if they have been convicted of a crime.

- nothing whatsoever to do with race, everything to do with (a) citizen status and (b) legally proven criminal behavior.

So: what you describe would be racist. What the article talks about that DHH was referring to clearly is not.

And please don't start with "dog whistle".

If you're going to accuse people of things, it does need to be them doing those things, not you interpreting.

If you can present clear evidence of him supporting actual ethnic cleansing, so not something that you somehow interpret as being equivalent, I am with you.

The return of the Transputer.

Put a Transputer in a Lego brick. Not "like a Lego brick", an actual Lego brick. Turn the notches into connectors for the serial link.

Plug'n'play.

So essentially React Native but with the code not left in TypeScript/JavaScript but transpiled to C++?

Which is why "data structures" probably also belong in the "variables" column.

> humans will eventually stop writing and reading code,

“Since FORTRAN should virtually eliminate coding and debugging…” -- FORTRAN report, 1954

http://www.softwarepreservation.org/projects/FORTRAN/BackusE...

Is it just me or are local models getting better (catching up) a lot faster than the frontier models are getting better (creating distance)?

If true, that would be a very welcome development.

Tried the app. Just gives me the error message "Failed to decode Homebrew JSON output".

   ]brew -v 
   Homebrew 7.0.1

So prescient.

Time for alternatives?

Sure!

To start:

An operating system is a collection of things that don’t fit into a language. There shouldn’t be one.

— Dan Ingalls, https://www.cs.virginia.edu/~evans/cs655/readings/smalltalk....

Let's start with an example, the Objective-Smalltalk equivalent of Python's simple web-server for a directory.

Python:

    python3 -m http.server 8000

This serves the current directory on port 8000. The invocation for the Objective-Smalltalk script doing the equivalent job looks as follow:

   ./simplefileserver.st ./ 8000 

Here is the Objective-Smalltalk 'simplefileserver.st' script:

   #!env st
   #-server: <ref>dir port: <int>port
   framework:ObjectiveHttpd load.
   dir asScheme waitOnPort: port.

If you look carefully, you'll notice that this doesn't actually say anything about the filesystem. It just expects an argument 'dir' that is a 'ref' and then goes from there.

Here are some other example of what a "ref" can be:

   ./simplefileserver.st https://objective.st 8000
   ./simplefileserver.st sftp://otherhost.local 8000
   ./simplefileserver.st env: 8000
   ./simplefileserver.st defaults: 8000
   ./simplefileserver.st class: 8000
   ./simplefileserver.st scheme: 8000

The first will make this server act like an http proxy, the second will serve a remote directory, the third will serve our environment variables, the fourth will serve the contents of our process's defaults database (a macOS/iOS thing), the fifth our classes and the sixth the contents of the scheme registry where all the other schemes listed so far are registered.

"Ref" is anything referencable by a polymorphic identifier, a slightly adapted version of a URI. And any ref can be turned into a store (aka "scheme-handler"). And a store can be literally anything that can store and retrieve data by key/index/name/path, from local variables via filesystems, databases all the way to the World Wide Web.

So you can see that you can get the same location transparency that Plan 9 gives you without requiring operating system support. Which comes in handy when you can't switch to a new operating system. What's more, the support extends to a lot of things even Plan 9 probably wouldn't consider a filesystem (local variables?). Last not least, this does not require accessing everything via POSIX API and a byte-serialized representation. Definitely not "least"!

You can access via a byte-serialized representation (as via the http server above) if you want, and you can even access via POSIX API (there is a generic fileserver that serves any of these stores / scheme handlers) if you really, really want. I am not at all sure why you would want that.

And you can, of course, also access this via any language that can talk to C as the libraries that implement the functionality are written in portable Objective-C. Objective-Smalltalk just provides the proper linguistic support.

I think there is a good case to be made for these things not to be mediated by the operating system by default.

In Objective-Smalltalk[1], I can access a file as follows:

   hello ← file:hello-world.txt 

This is structurally the same way I would access a local variable, environment variable, database, remote http server etc.

   hello ← https://example.com/hello-world.txt
   hello ← env:GREETING
   hello ← var:greeting         //   hello defaults to var:hello

etc.

And you can also introduce shortcuts

   scheme:greeter ← ref:https://example.com/ asScheme.
   hello ← greeter:hello-world.txt

Or

   scheme:greeter ← ref:file:./ asScheme
   hello ← greeter:hello-world.txt

Sending -asScheme to a reference is just a shorthand that actually constructs a composition[2] of a "path relative" store with the underlying store of the original reference. So the following two are identical:

   scheme:greeter ← ref:https://example.com/ asScheme.
   scheme:greeter ← #MPWRelativeScheme{ base: 'https://example/com' } → #MPWURLSchemeResolver{} 

This composition mechanism can be carried further with post-processing, so for example an img-scheme can be constructed by composing an image-decoder store with the previous store

   scheme:img ← #MPWImageDecoderStore{} → scheme:greeter 
   helloPic ← img:wave.png 

And so on and so forth, caching is also a nice example.

[1] https://objective.st

[2] https://dl.acm.org/doi/10.1145/3359591.3359729

Have a look at Objective-Smalltalk[1][2], with polymorphic identifiers[3] (all identifiers are URIs), storage combinators[4] (virtual filesystems on steroids), and polymorphic write streams[5] (streams everywhere).

Objective-Smalltalk basically provides the sorts of things you write about locally at the language level. You could also move specific instances behind an operating system boundary.

[1] https://objective.st

[2] https://dl.acm.org/doi/10.1145/3689492.3690052

[3] https://dl.acm.org/doi/10.1145/2508168.2508169

[4] https://2019.splashcon.org/details/splash-2019-Onward-papers...

[5] https://dl.acm.org/doi/10.1145/3359619.3359748

To me the "Europe now has sovereign access" line by the CEO is a bit of hype, though technically it is true if you view Europe geographically, as he seems to do.

Andøya Spaceport has been a place for suborbital/sounding rockets, this is the first time a rocket has been launched into orbit from there.

https://en.wikipedia.org/wiki/Andøya_Space

> I don't have answers to your other questions as I'd ask them too.

They were questions to you, as it is you making various weird claims.

1. Who is "they"? Isar Aerospace and Arianespace are distinct entities.

2. Isar Aerospace is the first fully private European company (a startup) to do this. Arianespace and the Ariane rockets have heavy state/ESA involvement.

3. It is the first launch to orbit from geographically European soil. French Guiana is politically part of France, but geographically obviously part of South America.

4. Why would everyone have to launch from the same place, in your esteemed opinion? Why are some US payloads launched from California, some from Florida, and soon some from Texas and Louisiana?

5. Are all launch locations equally suitable for all orbits? Are all orbits equally suitable for all missions?

6. Norway is a lot closer to the rest of Europe than French Guiana is.

7. Why was SpaceX newsworthy, particularly on a startup-centric site, when NASA already exists?

I doubt that HN user yostrovs is Trump.