Posts by Len

Hm, why are so many DrayTek routers stuck in a bootloop?

Are you sure they are the same worldwide? I remember a Draytek modem (admittedly a modem not just a router) that had firmwares ending in _bt if you were on OpenReach.

AI summaries turn real news into nonsense, BBC finds

We know LLMs are poor at summarising

I've posted this before and it's still relevant.

LLMs keep demonstrating that summarising is their weak spot. They can shorten but, because they're inherently stupid and they have no idea what they are doing, these "AI" implementations are unable to distinguish the important from the not important. And that's key for summarising.

When ChatGPT summarises, it actually does nothing of the kind.

AI worse than humans in every way at summarising information, government trial finds

AI chatbot startup founder, lawyer wife accused of ripping off investors in $60M fraud

Re: Fiscal Reporting

In Europe you could do that thanks to the PSD2 directive. Under PSD2 banks are required to allow their customers to give read-only access to their own bank accounts to other parties in the interest of reducing the stranglehold banks have over their customers and to increase innovation and competition in financial services.

But, as so often, the technology is the easy part. The problem is that bookkeepers/accountants/auditors are paid by the company whose books they keep or audit. They lack incentives to pick fights with the management of the hand that feeds them.

OpenZFS 2.3 is here, with RAID expansion and faster dedup

Re: Kitchen-Sink and Reliability

Running apps, containers, and VMs on your storage boxes? Why?

Those features (and therefore that code) is not part of the OpenZFS code. It's part of the appliances that some companies make and that are build around OpenZFS. OpenZFS is still only a storage subsystem.

Eutelsat OneWeb blames 366th day for 48-hour date disaster

Re: Laugh or cry?

This is very interesting reading: How the lore of New Year defeated the law of New Year – how the English state gave up on insisting the new year started on 25 March. Especially if that explains why the UK's tax year still starts 11 days later on 5 April.

Re: Rocket Science

Don't forget Excel's year 1900 bug that is still present in every version of Excel. Excel incorrectly assumes that the year 1900 is a leap year

I bet scientists doing historical time series are frequently caught out by this bug.

Apple called on to ditch AI headline summaries after BBC debacle

We know LLMs are poor at summarising

LLMs keep demonstrating that summarising is their weak spot. They can shorten but, because they are inherently stupid and they have no idea what they are doing, these "AI" implementations are unable to distinguish the important from the not important. And that's key for summarising.

When ChatGPT summarises, it actually does nothing of the kind.

AI worse than humans in every way at summarising information, government trial finds

UK ponders USB-C as common charging standard

Re: See also: UKCA marks on goods. That didn't last very long

It depends on the product type. Sometimes to carry the CE mark you "just" need to certify that your products follow regulation XYZ for that product range and to provide a named point of contact for any enquiries around recalls, blueprints, and legal cases up to X years after the product was last sold.

For some product categories (building materials for instance, some tools) where the risk is deemed greater, there's required CE testing process that needs to be followed and the outcomes certified. And that was indeed a problem immediately after Brexit as UK based testing labs were not certified any more to conduct CE conformity tests. Eventually some UK based CE testing labs opened branch offices in the EU (mainly in the Netherlands and Germany) to keep providing CE Conformity testing to their UK customer base.

Re: A waste of time and taxpayer money

There's even a name for the phenomenon of the EU setting global standards, it's called the Brussels Effect. I expect we can soon chalk up the effort to standardise USB-C for all small devices as another proof point of the Brussels Effect in action.

Google finally addresses those bizarre AI search results

Re: "limiting its reliance on info written by everyday users"

Funnily enough the suggestion to eat rocks was reproduced verbatim from a piece in The Onion: Geologists Recommend Eating At Least One Small Rock Per Day

How two brothers allegedly swiped $25M in a 12-second Ethereum heist

Re: Good example of "novel" risks for tyro/blinkered coders

As far as I'm aware one of the furthest advanced CBDC projects has not even decided whether it will use a Blockchain, so the (unlikely if you ask me, they'd probably opt for a, more modern, 3rd generation Blockchain) choice for Ethereum has definitely not been made yet.

I wouldn't be surprised if the Digital Euro project wouldn't choose to use a Blockchain at all. Using a Blockchain has some benefits that wouldn't apply in a CBDC while doing it without a Blockchain would make it faster, cheaper and better to handle the scale of transactions we'd be talking about.

Mastodon delays firm fix for link previews DDoSing sites

A side effect of success

This issue has always existed of course, but it wasn't until Mastodon exploded in popularity that it became a frequent problem. After Elon Musk started to take an axe to Twitter the Mastodon network gained more users in some months than it had gained in the six years prior.

What happens is essentially the 'Slashdot effect', or similar to sending out a bulk email with links to your website. You may have hundreds of thousands of users with their own clients behind their unique IPs that all want a bit of data from your server, it's truly 'Distributed'.

I make sure that all the content that Mastodon servers would want from my sites to generate a link preview is static and that mostly seems to prevent the issue, for me at least. But, if you have a site that needs to fetch OpenGraph snippets from a database you need to take optimisation pretty seriously, or use a CDN proxy type services to do some of the caching for you.

Some scientists can't stop using AI to write research papers

Or the explosion of the use of the word "delve" since the release of ChatGPT

There's also the outright explosion of the occurrence of the word "delve" in PubMed articles that coincides neatly with the release of ChatGPT

Firefox 124 brings more slick moves for Mac and Android

Consent-O-Matic add-on

Sadly, a feature of the beta that's missing from the final release is the Cookie Banner Blocker, although users in some regions may get to enjoy this when browsing in Private mode.

I have been using the Consent-O-Matic extension by the University of Aarhus to get the same functionality, It recognises many of the typical cookie banner formats and automatically opts-out from as much as possible. I can definitely recommend it.

International effort to disrupt cybercrime moves into operational phase

Re: "brought cybercrime to the forefront of discussion among CEOs and boards of directors"

Agreed, you could halve the severity of the punishment, if simultaneously you double the chances of getting caught you've just created a greater deterrent.

FOSS replacement for Partition Magic, Gparted 1.6 is here to save your data

Re: because it's licensed differently from the kernel

I don't think Oracle has a ZFS implementation that can run on Linux. Oracle inherited Solaris when it purchased SUN Microsystems and that came with a fairly old version of ZFS. That version of ZFS was ancient by today's standards and only ran on Unix. Oracle has done some work on their own ZFS implementation but relatively little compared to a number of other players in this space (such as Delphix and Nexenta Systems for instance). Rumour has it that the entire division that used to work on Oracle ZFS was closed a few years ago.

The developers behind Illumos, iXSystems, FreeBSD, Nexenta, ZFS-on-Linux and Delphix all pooled their ZFS code a few years ago under the OpenZFS banner. That's why OpenZFS is so far ahead of Oracle's fork of the original SUN code.

If Oracle wanted a ZFS implementation that could run on Linux they would most likely have to rely on the OpenZFS code. And somehow I doubt they would do such a thing.

Re: Must give it a go some time

As Liam says, one shouldn't use GParted to convert or resize your ZFS volumes. But it is vitally important that GParted is ZFS aware and recognises existing ZFS partitions so it doesn't accidentally damage anything. Fortunately GParted knows about ZFS and can detect, move and copy ZFS sections.

One could wish that GParted could do more with ZFS but it's probably not wise for them to take that on. ZFS (at least the OpenZFS implementation of ZFS) development is very active and moves fast. The GParted developers would for ever be playing catch up with new features of OpenZFS and that is risky for something as vital as a filesystem. It's safest to stick to the official OpenZFS tools, they are free, open source and multiplatform so I see no reason why someone wouldn't just use those.

Web archive user's $14k BigQuery bill shock after running queries on 'free' dataset

Slappable jerk

Do you know the 'Slappable jerk' character 'the average Redditor'? He is scarily accurate, and very slappable indeed.

Crowning glory of GOV.UK websites updated, sparking frontend upgrades

Re: If they *really* want to improve the experience....

That's nothing. The only reasons I still have a fax is because I some times deal with American organisations and only a few years ago I had to fax a form and I could only fax it during (their) office hours while on the phone to the recipient because they had to walk to the fax machine to pick it up and confirm receipt. Someone needs to tell the US about this invention called the Internet.

Aircraft rivet hole issues cause delays to Boeing 737 Max deliveries

Re: And then there's the engine inlet problem...

Doesn't Airbus have a "break' in their cockpit commonality in that the A350 and A380 are in a class together, and the rest of the A3xx are in a class together? Presumably the A220 is also it's own class because it was originally developed by Bombardier and I doubt Airbus had the chance to completely switch cockpit to an Airbus cockpit so late in the development.

Re: Boeing should

There's an old story that Jaguars didn't become reliable until Jaguar's management was forced to only drive Jaguar.

Yes, in a way we should be seeing more of these come up throughout the year. It should be expected and could be a sign that they are trying to turn things around.

As bad as the drip feed of bad news might feel, if it amounts to things that are caught early enough they must hope that the people that matter (FAA and EASA, aircraft customers) see this as a positive. Boeing is fortunate that airlines have high switching costs to go to the only real competitor and that passengers have limited say in which aircraft they fly with.

Firefox 122 gets even more competitive with Chrome on translation

Re: common parlance

"How does the Heisenberg compensator work?" Just fine, thank you.

Wait, hold on, everyone – Mozilla thinks Apple, Google, Microsoft should play fair

I worry about Apple letting go of the WebKit requirement

While morally, and possibly legally, dubious I do worry about Apple dropping its WebKit requirement on iOS.

IOS has an enormous installed base, probably about 1.5 billion users. That is 1.5 billion people that use the WebKit render engine instead of Google's render engine called Blink. Currently even the laziest web designer will need to make sure that the sites they produce work both on WebKit and Blink.

If Apple drops that WebKit requirement we won't be entering a brave new world where smaller engines such as Mozilla's engine called Gecko suddenly stand a chance. We will probably see 1.3 billion WebKit users move to a browser that uses Blink. Lazy web developers could just say that their sites works fine in Chrome so you should just install Chrome. It will only increase Chrome's dominance in the market.

Who is to say that Google will stick to interop improvement attempts between render engines? I wouldn't put Google in a top 1000 of trustworthy or reliable tech giants and personally believe it would be the end of the open web.

Re: Amazed that FF isn't used more

I am very sceptical about browser usage statistics and always have been.

Firstly, I'm not aware of any serious methodology to measure this. There are things such as Statcounter and NetMarketshare but their methodology is a joke. They sample a tiny amount of (mainly American) websites and then extrapolate, assume, and extrapolate a bit further to get a sense of what's happening on Argentinian or Zimbabwean websites.

Secondly, some browsers (such as Firefox) are stricter in what additional JavaScript or cookies (that is not necessary for the site to appear) they allow to run. I use Firefox and when I visit some sites where I have access to Google Analytics my visits don't show up.

Thirdly, ad blockers prevent the loading of many measurement scripts and most block Google Analytics by default.

The above points would have been less relevant if their distribution was equal over various countries and browsers but it's not. People who use Chrome either don't care about privacy or don't know about it. People who use Firefox are much more likely to care about those things. That skews most measurement tools against Firefox. The only way to get a good sense of browser usage would be to make a globally representative list of the top 1 million websites, ask every single one to share their server logs (Unlike, let's say Google Analytics, server logs don't lie) and use that to get some sense of browsers used to surf the web (though you'd still need to filter out all the bots that lie about what browser they are). As this is unrealistic I don't think we'll ever have a reliable figure for browser usage.

Add to that that the total size of the market has exploded. When Firefox had a market share of 25% that will have been 25% of 1 billion internet users so about 250 million users. If there are currently 5 billion internet users and Firefox has a market share of 3% that would mean about 150 millions users. Considering Firefox own stats report around 200 million users (based on browser checks to see if there is a new version) the decline in absolute terms hasn't been as big as it may appear.

Now, this doesn't mean there is no problem. If website operators make sites that are poorly accessible for Firefox users, or worse, they only look at the website stats that Google Analytics (probably one of the world's worst website analytics tools because it's blocked by almost every privacy plugin and ad blocker) they will think their sites are not visited by anyone using Firefox.

Researchers confirm what we already knew: Google results really are getting worse

Re: Qwant search engine

I believe the challenge that DuckDuckGo has is that it's very reliant on Bing's data for its operations. That is why I advise clients who wish to be found by search engines to not just submit their site to Google but also to Bing because you actually feed into a long tail of search engines that also (at least partially) rely on Bing's data.

If Bing's data is not as good as it once was (and there appears to be serious indications that they have declined just like Google Search) then that will ultimately also impact DDG.

Re: Not just lower quality..

I can’t find the link now but it wasn’t too long ago that I read that there are research projects that track the size of the corpus that Google and Bing use. They had detected a significant drop in the number of sites that are actually indexed (I seem to recall that it was only in the tens of thousands of sites but that can’t be right) as they seem to actively get rid of a whole chunk of the database.

There was the incident not that long ago of Techdirt.com, not a small site, just disappearing from Google altogether. Not pushed back to page 37 or summat but removed from the corpus entirely.

It’s almost as if they’ve changed strategy and no longer want to provide the best Search but rather the most profitable Search.

Re: The only real test for AI

I think the problem is that a lot of the crap in search results is produced by AI (or actually not AI but LLMs).

Europe's monopoly cops suddenly rather curious about Microsoft's $13B for OpenAI

Under the radar takeover

The big challenge is that MS doesn't have formal majority control of OpenAI through share ownership (which would make it subject to some M&A and competition provisions) but that it effectively has near full control of OpenAI. It will be interesting to see if this means that MS can stay out of legal trouble as current competition law may not be sufficient to do something about it.

Nearly 200 Boeing 737 MAX 9 airplanes grounded after door plug flies off mid-flight

Re: Numbers

The European Union Aviation Safety Agency (EASA) hasn't grounded any 737s because apparently these specific types aren't used in Europe. That suggests not all 737s use this design.

Open source PostgreSQL named DBMS of the year by DB-Engines

Re: SQLit

I've only used SQLite in some small to medium sized web projects but as I understand not all concurrency is the same when it comes to SQLite. Read concurrency is high because it doesn't require locks, hence the benchmarks that suggest that having 100,000 visitors a day to a website running off SQLite is fine. If the server can keep the entire DB in its RAM it's even blazing fast, which probably explains why SQLite is sometimes faster than having a DB server because it doesn't have to open connections.

The challenge is that only one user can write at any time due to file locks. That is fine for an average website CMS where you rarely have more than a handful of people adding things roughly around the same time. It will be a problem if the visitors to your site do more than read content and you need to allow writes to the DB too.

As of a couple of years my modus operandi is to start projects using SQLite. If the projects take off and run into limitations I upgrade to Postgres. Talking to an SQLite DB uses more or less a subset of Postgres anyway so migration is easy. I haven't touched another DB system in years.

Google to start third-party cookie cull for 30 million Chrome users

Re: So….

I suspect the reason that Google is happy to cut off some data gathering exercises by third parties is that they obviously won't be cutting off their own data gathering exercises that are hardcoded into Chrome.

Re: Only stupid people use google crome

Agreed, if you read this far (in other words, you're reading The Register) you should not be using Chrome.

Apple pops blue bubbles of Beeper Mini's iMessage service again

Also an anti-SPAM measure

I suspect that one of the reasons for Apple fixing these holes is an anti-SPAM measure. If a third-party can gain access to Apple's servers to deliver messages what's to stop a SPAM service to dump messages into the iMessage network? So far my iMessage has been pretty free from SPAM, as opposed to my WhatsApp or Telegram accounts.

Fired OpenAI boss Sam Altman may join Microsoft

Microsoft's investment was mainly in kind

As I understand it the majority of Microsoft's investment was not in form of a transfer of money but in the form of donated computing capacity at Azure. Even if OpenAI were to tank, MS wouldn't have lost that investment.

What happens now is interesting. I don't get the impression Altman was the brains behind OpenAI, he was the smooth talking figure head (who converted the existing non-profit into an investment vehicle), not the scientific brain. It will depend on how many scientists he can persuade him to follow him to MS whether it has a chance of taking off. Meanwhile MS could use access to its computing capacity to put pressure on OpenAI.

Excel recruitment time bomb makes top trainee doctors 'unappointable'

EuSpRIG horror list

This is a nice addition to the European Spreadsheet Risk Interest Group Horror List.

Microsoft drops official support for Python 3.7 in Visual Studio Code

Re: Python 3.11 is where you want to be

Interesting about Facebook's use of Python, I wasn't aware of it. I know Facebook was originally created in PHP but they ran into PHP's limits and had first created their own PHP server and then tweaked the language itself to suit their needs.

There was a period where moving from PHP to Django (the Python based web framework) was the logical step up so perhaps Django is where they went after PHP.

Musk's first year as Twitter's Dear Leader is nigh

Re: Mastodon remains the most exciting alternative

Have you checked this list? It’s a giant list of lists of scientific disciplines and academic communities on Mastodon. Well worth a check for accounts to follow.

Academics on Mastodon

No joke: Cloudflare takes aim at Google Fonts with ROFL

GDPR compliant alternative to Google Fonts

For those looking for a GDPR compliant alternative to Google Fonts I can recommend fonts.bunny.net. They have drop in replacements for all Google fonts and even a WordPress plugin that redirects it for you if can't (or prefer not to) change your WP themes with hardcoded links to Google Fonts.

If anyone finds an $80M F-35 stealth fighter, please call the Pentagon

When I grew up the joke was that you only ever saw artist impressions of the F-16 because they tended to crash before someone could take a photo.

OpenAI's ChatGPT has a left wing bias – at times

Re: What do you mean by "Liberal"?

Are you sure you're not confusing the FDP with AFD and VVD with PVV? Because AFD and PVV have some fascist tendencies but VVD and FDP certainly don't. The VVD and FDP are pretty mainstream centre-right parties. You might call them uncaring, greedy, catering to a narrow base of people with above average wealth who've 'made it' while neglecting anyone else but you can't call them extremists.

Re: What do you mean by "Liberal"?

Good point. The US has quite a unique definition of the political label "Liberal". Most Liberal parties all over the world (from the German FDP and the Dutch VVD to the Australian Liberal Party or the UK's 19th century Liberal Party) are right-of-centre parties focusing on small government and low taxes, it's only the US where it seems to mean almost the opposite.

OpenZFS 2.2 is nearly here, and ZFSBootMenu 2.2 already is

Also in TrueNAS

The first beta of TrueNAS SCALE 23.10 also includes OpenZFS 2.2, though it's a beta and based on OpenZFS 2.2.0 RC3 so I would only use it for testing purposes at the moment.

iXSystems, the company behind TrueNAS is a major contributor to the OpenZFS code and quite a few of the improvements in the latest OpenZFS are either done by iXSystems developers or by others but sponsored by iXSystems. I believe they were also behind getting the RAID expansion development back on track.

Not call: Open source gurus urge you to dump Zoom

Re: Use Jitsi

That is a very good point. Fortunately the Jitsi software is open source and can be self-hosted. That means you could host your own, or choose an EU-based, GDPR-compliant, Jitsi service. I've just found this one for instance: Fairkom.eu.

Zoom Out

They missed a trick in not calling the campaign Zoom Out. It works on two levels, pushing Zoom out of the building and zooming out to see the alternative services.

Also, this excellent piece of sleuthing suggests it was far worse than just some unfortunate wording: Zoom knots itself a legal tangle over use of customer data for training AI models. It seems clearly in breach of many European privacy safeguards and, for instance, I don't think any judge would agree with their definition of opt-in ("we start using your data from the start and then notify you so you can leave the call if you don't like that" is not opt-in). Furthermore they either seem to accidentally or on purpose make a complete hash of their understanding of the GDPR, or who their regulator is.

India launches contest to build homegrown web browser

Re: I wish them luck

I seriously hope that this will be based off the Firefox code (who knows, Mozilla might even help them because at some level it's in their interest) and not just another slightly adapted version of Chrome with its Blink engine.

It sounds counter-intuitive at first but I seriously worry that Apple might be barred from requiring the use of the WebKit engine as underlying render technology on iOS. Such a ban may appear to create more user choice but it actually would reduce choice because it would just allow companies to make sites that only work on Blink and if you complain they'll just tell you to install Chrome because "it works just fine when they check it on Chrome".

Firefox currently has about 200 million users making its Gecko engine the third most used engine to browse websites. If the "IndiaBrowser" would use the Gecko engine too it could easily double the use of Gecko worldwide, making the engine competition more of a three horse race again, instead of one horse on wheels with rocketboosters while firing at the competition and two horses behind it dodging bullets from a bully.

If India is seriously worried about the dominance of Western tech then not using Chrome/Chromium/Blink, which is effectively a trojan horse pushed by Google, should be a good first argument to pick something else.

_{I don't see building another engine from scratch as a viable alternative. Developing browser engines has become so complex now browsers are essentially application platforms that the complexity is probably in the same ball park as building an operating system. I would argue that developing a kernel from scratch is easier than building a web browser from scratch.}

How to spot OpenAI's crawler bot and stop it slurping sites for training data

Re: The risk with Robots.txt

I get that, thanks. But I was wondering how to do it selectively. So, let's say I want ChatGPT to hoover up data from directory X but not from directory Y, without telling all and sundry that directory Y exists so I can't do disallow Y in Robots.txt

The risk with Robots.txt

Can someone enlighten me?

Robots.txt has been used for/against search engine bots for ages. It has always come with the warning that if you don't want people/bots/crawlers to know that directory B exists you should not explicitly allow crawling of directory A while explicitly blocking crawling of directory B. You're asking not to look somewhere and only scrupulous bots would honour that. The solution is usually to sort it out at page level so a page you don't want to be crawled has a meta tag blocking it.

How would one do this with the ChatGPT bot? Will it also look at the page meta tags? I don't trust some makers of AI bots to not use a block attempt to explicitly go and harvest data from directories that I have disallowed. It might just give them an edge over OpenAI.

How could I let ChatGPT freely crawl my FAQ or About Us page but not my Content page?

Aspiration to deploy new UK nuclear reactor every year a 'wish', not a plan

Re: Technical marvel, but it's the economics, stupid

You are looking at consumer cost, which is hard to compare and doesn't say much about production cost. Some countries have amazing grids, some country's grids are held together by tie wraps and duct tape. Some countries add extra levies to pay for grid upgrades, some don't. Some countries add extra levies for an energy transition, some don't. Some countries have a well functioning utilities market, some don't.

If you want to look at the wholesale cost per source you'd need to look at what's called the 'Levelized cost of electricity' (LCOE). Then you'll typically see, from cheapest to more expensive, something along the lines of Large-scale solar -> offshore wind -> onshore wind -> combined cycle gas turbine (CCGT) -> geothermal -> nuclear -> small scale solar * -> coal (hard) -> coal (soft) -> more obscure sources such as old fashioned gas plants, oil, etc.

Occasionally offshore wind and onshore wind trade places in that order, dependent on geography. Spain will probably have cheaper onshore than offshore wind whereas Denmark will have it the other way around. Also, load factors vary per country so some countries will have wind farms that produce about 60% of the time, other countries only 30% (often, but not solely, based on whether they have more offshore wind parks as offshore has a higher LF). But interestingly enough, even an abundance of coal in a country doesn't reduce the wholesale price that much. And neither does having serious economies of scale in nuclear, like the French have.

The thing is that the fuel cost only plays a part role in the total cost of producing a MWh. Obviously fuel cost for wind and solar plants is zero per MWh whereas gas or uranium or coal cost money but the total cost of construction, operation, maintenance, decommissioning etc. is where the real expenditure lies.

* _{Small scale solar, such as people putting a few solar panels on their house, is not very cheap in the grand scheme of things but because it's individuals who become less dependent on their energy supplier it might still make economic sense for them. Even more so because most other forms of electricity production are usually beyond the reach of individuals.}

Technical marvel, but it's the economics, stupid

Nuclear power plants are without a doubt a technical marvel. Its designs, its complexity, its processes, its safety procedures are truly part of a great human achievement.

However, the problem the nuclear power industry has is that the economics, not the technology, make it outdated. It simply cannot compete on price because construction is so hideously expensive. There was a time private power companies had their own nuclear power plants, then came a time that only governments had enough money to finance them. Now were at a time where even many governments don't have the required funds any more to make it happen. Every time there is a nuclear power incident somewhere on the planet (on average every fifteen years I believe) the cost for new reactors shoots up yet again.

The UK had to get China to fund its latest reactor and could only get the Chinese to sign the contract in return for a strike price twice the current wholesale electricity price, guaranteed for 30 years! While the trend on the wholesale electric is only downward. That means that in ten years we'd probably be paying the Chinese four times the wholesale price, in twenty years ten times the wholesale price. It means that nuclear power plants are essentially running on subsidies to produce reliable baseload. It's financial madness to overpay so much just for reliability and it opens up the economics for alternative systems of reliability (batteries of all kinds for instance) to grossly undercut nuclear power.

Even France, easily the world champion in using nuclear power to fulfil its demand, is investing in wind power like it's going out of fashion because the economics of nuclear are starting to hold them back.

Nuclear is on course to drop out of the top five cheapest ways to produce electricity and I see no way it can compete with, for instance, solar and wind power which, all things considered, are becoming insanely cheap. I don't see how nuclear could ever become cheaper than those two again. There's something to be said for nuclear to power submarines, perhaps even spaceships, but not countries any more. I think that ship has sailed.