Frank Zimper 🕯️🐘

@fzimper@bildung.social
372 Followers
1,094 Following
5.7K Posts
Ehem. Lehrer (Inf, M, Ph) in BW. #FediLZ
Diverseshttps://www.frankzimper.de/

Wecker ausschalten.

#ServiceToot

Digitalisierung, die beim Nutzenden ein Smartphone voraussetzt, ist Enshitification.

Warning, long text

Power Outage in Spain – An Analysis

Solar energy comes out of your panels as direct current (DC). That’s all well and good, but homes and grids run on alternating current (AC). Enter the inverter – the humble box that turns solar wizardry into household juice.

Now, inverters aren’t just fancy plug adapters. They have to sync up with the grid – which means they generate exactly the same frequency as the rest of the system. No grid? No syncing. In that case, the inverter goes into what’s called island mode and produces power only for local use. So, if my solar system isn’t connected to the external grid, it can’t run the house – but it can still power two little emergency sockets. Cheers, I guess.

Normally, the grid runs at 50 Hz – that’s hertz, not some obscure Scandinavian metal band. But this frequency can wobble a bit. Physically and technically speaking, it rises when there’s too much power and not enough consumption, and falls when there’s a hungry grid and not enough electricity to feed it.

To keep the grid safe, inverters have an emergency shutdown feature: if the frequency goes over a set limit (apparently around 50.2 Hz), they also jump ship and go into island mode.

Spain’s energy mix is a bit unusual: lots of nuclear, lots of renewables – and a large chunk of those renewables are solar. Makes perfect sense in a country where “cloudy” means three fluffy cotton balls drifted by.

Now, nuclear energy comes with two charming quirks. First, you can’t change its output quickly – it’s not a dimmer switch, more like a cruise ship rudder. Second, nuclear plants cost nearly the same to run at half speed as they do at full throttle. So, naturally, you want to keep them purring along at max capacity.

Then came Monday, with weather conditions perfect enough to make a solar engineer weep with joy: loads of sun, plenty of wind. By 9 a.m., Spain’s energy needs were entirely met by nuclear and renewables. In fact, they had surplus electricity and began exporting it by the bucketload. They shut down everything easy to shut down – but nuclear? No chance. It stayed full steam ahead.

Then, two unfortunate things happened: one transmission line to France caught fire (as you do), and another developed resonances due to meteorological oddities.

So far, this is all well documented. Now we step into speculation territory.

These instabilities meant Spain couldn’t get rid of its excess electricity. The grid frequency rose past that critical 50.2 Hz mark – and boom: many solar systems switched to island mode. At that moment, they were providing nearly 15 gigawatts – around 60% of the national supply. And just like that, poof – they were gone.

Suddenly, two-thirds of the electricity vanished. Wind, nukes, and batteries couldn’t keep up – quite the opposite, in fact. To prevent damage, the nuclear plants initiated emergency shutdowns. Not great. (More on why that’s bad in a bit.) Within seconds, the entire grid collapsed. The solar systems were poised to help – but there was no grid left to sync with.

Everything went dark.

Portugal and southern France were also knocked offline, as they’d been happily sipping from Spain’s excess power. The European grid wasn’t amused and unceremoniously kicked Spain out of the club. France, with a bit of backup and a stiff upper lip, restored its network fairly quickly. My home automation system even picked up the moment the frequency dipped and France cranked up its own generation.

Portugal got the rough end of the stick. With fewer reserves and being smaller in size, they couldn’t help themselves – and no one else could help either, since Spain’s their only neighbour.

Rebooting the Grid – Why It’s a Right Pain

Restarting a collapsed grid isn’t just a matter of flipping a giant switch. It’s tricky for two reasons:

  • Generation and consumption have to be in perfect balance. If not, we’re back to square one.
  • Nuclear power plants can’t just be turned back on. After an emergency shutdown, they suffer from something called xenon poisoning (yes, one of the very same issues that made Chernobyl a household name). You’ve got to wait for that to wear off – which means the reactors were still offline two days later.

The fix? You split the grid into smaller bits. For each chunk, you build up some capacity, bring it online, then move on to the next. Rinse and repeat. This takes hours. Meanwhile, the sun moves across the sky – and even if you do reconnect the solar arrays, they won’t produce nearly as much as before. Come 8 p.m., they’re more or less useless.

So Spain needed outside help. They were gradually reconnected to the European grid – in small, careful steps. Without that assistance, large parts of Spain would probably still be in the dark. That’s why electricity came back first in places like Barcelona, close to the French border, while Portugal endured the longest wait.

Notes & Musings

  • Considering the scale of the event, the recovery was impressively quick. In San Sebastian, power was back within 2 hours. (For comparison: Wismar in Germany had a 45-minute outage last year because one substation had a wobble.) Portugal got its power back after 23 hours. I had expected one to two days.
  • This was the largest blackout in Europe in 40 years. If, as suspected, climate-related factors helped spark (pun intended) the situation, then modernising the grid to better handle volatility is absolutely essential. That includes implementing the long-debated power zones in Germany.

Have you seen Episode one of the seventh series of Black Mirror?

* spoiler ahead *

A dystopian piece about an implant that's essential to survival but becomes obsolete over time. The company provides a - costly - upgrade which is not affordable to the couple who need it.

Luckily it's just fiction. Right? IT IS FICTION, ISN'T IT???

Well...

https://www.sapiens.org/culture/planned-obsolescence-cochlear-implants/

Wikipedia page about the #BlackMirror episode:

https://en.wikipedia.org/wiki/Common_People_(Black_Mirror)

#CochleaImplant #plannedObsolence

Who Pays the Price When Cochlear Implants Go Obsolete?

Some cochlear implant users who can’t afford to keep up with compulsory technology upgrades are now losing their hearing.

SAPIENS

#RÜCKRUF | #WARNUNG | Aus aktuellem Anlass: Verbraucherinformation von ROSE BIKES (SCRUB Mountainbike > Unregelmäßigkeit an Schweißnaht des Steuerrohrs kann zum Reißen dieser Schweißnaht führen = STURZGEFAHR / UNFALLGEFAHR)
https://www.rosebikes.de/sicherheitshinweise-und-r%C3%BCckrufaktionen

Siehe auch: https://www.produktrueckrufe.de

Sicherheitshinweise und RĂźckrufaktionen Alpina | Entdecke die Welt von ROSE Bikes

Sicherheitshinweise und Rückrufaktionen Alpina – ROSE Bikes bringt Menschen aufs Rad • Qualität und Innovation seit 1907 • Online und in deiner Nähe

Reminder: The remaining number of registrations is available tomorrow, starting from 14:00 CEST!

Finally had time to watch Pewdiepie's "I installed Linux" video.

Don't know what I was expecting... but *wasn't* expecting a great list of reasons why, in 2025, you should stop using Windows and install Linux—he even mentions how to use systemd-analyze!

https://www.youtube.com/watch?v=pVI_smLgTY0

I installed Linux (so should you)

YouTube
Ihr Unternehmen steht also fĂźr Ăźberteuerte Preise, schlechte Beratung und miesen Service? Willkommen in der Bundesregierung!

Ich habe schon darauf gewartet - einige hier sicherlich auch:

„Männer, die die Welt verbrennen“ von @chrisstoecker

ist seit heute bei der Bundeszentrale für politische Bildung erhältlich! Meine Bestellung ist gerade raus…

#Klimakrise #Zukunft #Nachhaltigkeit

Hier sieht man im 48-Stunden-Zeitraffer den Mond gelegentlich durchs Bild flitzen.
×

Maybe we are on the cusp of AGI…

#agrokinference #ai #agi #naturalstupidity

@bsletten satire requires higher order thinking and is not for the gullible. GenAI is a great example of the Dunning Kruger syndrome 🤣🤣 #genai

@dalonso Here's alt text for the image:

A newspaper clipping shows a text article about a mathematician, Timothy Gowers, who jokingly tweeted that an AI model solved a math problem. The article discusses how this led to a false, AI-generated search result claiming the AI solved a well-known problem in graph theory.

@bsletten Do you have the source?
@usernaut No, it was posted without on another network. I would have included it if I knew. I’ll see if I can find it.
@bsletten @usernaut Given the font, the source is Private Eye.
AIprl Fool

In Private Eye this morning, I read about a nice April Fool joke perpetrated by Tim Gowers. He claimed on X that he had succeeded in getting Grok to solve the “well-known Dubnovy Blazen probl…

Peter Cameron's Blog
@bsletten Maybe at some point, people will start seeing AI as so wrong that no one will trust it anymore. And then the bubble will burst
@altf4 that's not how this will turn out. I'm afraid truth will burst before AI does.
@bsletten "Dubnový Blázen" 😭 Did AI also translated that?

@bsletten This is a great practical joke, but please be careful not to confabulate popular machine learning models with *all* of #AI.

For example, logic theorem proving is the oldest, most mature, and most mathematically rigorous of the subfields of AI. #Logic and #mathematics are formally equivalent. The “Logic Theorist” in 1956, programmed by Newell, Simon and Shaw was arguably the first computational system to perform deduction using propositional #logic. It managed to prove 38 of the first 52 theorems of the “Principia Mathematica” (1927) by Whitehead and Russell. This breakthrough was a major impetus for the study of #ArtificialIntelligence itself.

Automated theorem provers are used today to check the correctness of computer circuit design, formal specifications, software correctness, and serve as assistants to mathematicians for analyzing complex proofs.

@meltedcheese Please be careful not to confabulate everyone who isn’t you as needing a lecture.
@bsletten omg I would like to frame this reply.
@artcollisions @bsletten It would be a reasonable thing to do.

@bsletten @meltedcheese

Well I for one found his reply interesting. 

@meltedcheese @bsletten Do you have an example of your last statement? Recently formal proof check using systems such as Roc or Lean have gained traction, but they are not automated provers: human encode proofs, the machine checks its correctness.
@BrKloeckner @meltedcheese @bsletten They are partially automated provers. You can for example write a tactic that will automatically solve an equation that can be solved by only using associativity and commutativity.
An actual example is linarith, a tactic that can solve linear arithmetic problems :
https://gebner.github.io/mathlib_docs/tactic/linarith/frontend.html
mathlib docs: tactic.​linarith.​frontend

@meltedcheese this was really interesting; something I sort-of knew but hadn't threaded together. Thank you.
@meltedcheese @bsletten
yeah yeah we know that computers can help us with math and it’s awesome but can we concentrate on fighting what the Tech industry and everyone else is calling “IA” because it’s destroying human civilization, instead of arguing about semantics?

@bsletten

Call me naive but folks who intentionally poison #AI and then celebrate confidently incorrect answers are....

... stalwart defenders of humanity and not at all like the third grades who draw dicks with a pen, in the school library encyclopaedia !

@bsletten Here is how Deepseek sees it ... weird.
@bsletten where was this printed?
@anselmschueler It appears to have originally been in Private Eye Magazine. I just saw it referenced w/o citation.
@bsletten @BeTongLen I am finding it helpful to refer to any AI instance as Mr. Magoo.
@bsletten it looks like it's fixed?
@sharlatan By hand, I’m sure.
@bsletten Long before we attain AGI, we'll have attained AGS.

@bsletten

Let's rename AI 'trump computing'. He'd probably take it as a complement.

@bsletten going above human level by making human stupider

@bsletten #alt4you Part 1 of 2, photo of what appears a print news article:

"AFTER several unsuccessful attempts," the mathematician Timothy Gowers posted on Twitter on 1 April, "I found a prompt that got Grok to solve a maths problem (the well-known Dubnovy Blazen problem in graph theory) I've been working on for over a year. How long till it's better than human mathematicians across the board?"

@bsletten Alt text part 2 of 2:

This was, of course, a joke, signalled by Dubnovy BlĂĄzen being Czech for "April Fools" Several days later, however, others noticed that running a Google query for "Dubnovy Blazen problem in graph theory" resulted in an Al-generated search result that confidently proclaimed that it was "a well-known problem that Al model Grok solved after several attempts". It's wonderful watching the well of knowledge being poisoned in real time.

@cwicseolfor Thank you. I will address. I certainly meant to.
@bsletten It deserves the reach! (Also, thanks!)
@bsletten while the humor of this incident is not lost on me, I was curious to see what another LLM might provide in response to “What is the Dubnovy Blazen problem in graph theory?” I’m not a mathematician or graph theorist, but I speculate these outputs are possibly hallucinatory? 💭🤔
Jonathan Oppenheim (@postquantum.bsky.social)

Tim Gowers @wtgowers.bsky.social claimed that Grok solved the DubnovĂ˝ BlĂĄzen problem as an April Fools Day joke. Now Google thinks the DubnovĂ˝ BlĂĄzen problem is a real problem in graph theory (it's Czech for "April Fools").

Bluesky Social
@bsletten Well that's a bit ironic, isn't it... "dubnovy blazen" is absolutely not april fools in (my native) Czech. It's an overly literal translation of the English term and doesn't exist in my language where we say simply "aprĂ­l" (archaic word for the month of April) or "aprĂ­lovĂ˝ Ĺžert/ĹžertĂ­k" (April joke, again a bit archaic). So a mathematician, while trying for a gotcha with a BS generator, produced BS that ignores cultural context... but is now out there, polluting our information space.
@animalculum @bsletten wait…you‘re actually blaming the mathematician?
@flq @bsletten as far as I understand it, TG - the quoted mathematician - was the one who invented the name of that fake maths problem, to make a joke about "AI" making shit up. In doing so, he used my language as a plaything without bothering to do a simple research (as anglophones tend to do). And then, people started reporting on it, as if he did the translation right, not checking anything. So yeah, I'm also blaming the mathematician. "AI" being BS is a given. Am I missing something?
@animalculum @flq @bsletten I want to try and explain why you shouldn't blame the mathematician but I can't stop laughing😅
@dilmandila @animalculum @flq @bsletten He's not "blaming" the mathmetician. You're worried about "blame" - you're missing the point. Humans poison the well of information all on their own. Case in point, here we are reading another post on this story and this time it is in image form - we are replicating Gowers' original lie.

@jenzi What a mess! Then again, carelessness is humanity’s middle name.

But please enlighten me on Gower‘s lie.

@flq the translation is wrong. Which you know.
@flq You clearly didn't read the post you replied to, or any since, and are saying AI is the issue. You comment on human carelessness while not trying yourself. This is slop.

The inability of LLMs to tell true and false apart must be closely related to another of their failings, the odd stylistic lack of irony in their writings. Real living writers use irony all the time, and real living readers understand it just fine.

@bsletten

#llm #ai

@bsletten yeah, but that only works for terms that either do not exist like DubnovĂ˝ BlĂĄzen, or they have so few referents that's equivalent to not existing. I wish we could poison these things so easily for everything. Maybe we should do a webring of just repeated falsehoods, maybe even shared via torrent or plain rsync mirrors: one master editable source so it's easy to add crap and millions of copies everywhere.
@bsletten Anyone can be caught with an April Fool's day joke. I know I have been. But when I realised my mistake I have laughed at my gullibility and then corrected my information. The real question is: Will AI correct itself of its own accord without human intervention or will that response live forever in the data sphere?
@bsletten This isn't the conclusion I would draw. It's more like we're on the cusp of a total loss of signifiers.

@bsletten

Very good math problem example of the real AI , "Artificial Ignorance"

@bsletten I am just wondering which AI they used for the translation. 😅
Artificial General Stupidity?