rob_c

_sigh_ 2020, 2021.... I'm talking about clearly 2023 onwards once it was crystal clear to everyone fluent in actual stats what the risks were.

Again, give me the ~15million bodies we expect to see dead in India for a 1% fatality vs the ~100k that is a -serious- discrepancy from a nation that hasn't mastered other hygiene issues yet at a national level. And I'm supposed to believe they magically conquered cv19 better than the crazed lockdown nations of the west?

simply voting against brexit helped save gran remember: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3955345

Oh please don't give me the "le multiple studies in bio" argument. By that logic magic mushrooms and witchcraft are far more reliable due to 'working' for centuries. The field is full of reproducibility crises and follow the leader herd mentality, chemistry, even bio-informatics maybe, but bio as a whole, eesh...

Collecting the data in controlled conditions is exactly my point, you're trying to infer to a whole general populous on something that became insanely politically charged. Given the complete absence of admitting this or even trying to factor this into studies makes the whole thing a stack of cards.

and yet they still do. by that logic move away and stop them visiting. conclusion drawn from your arguments...

I'm not blaming victims, I'm blaming those in charge of helping the victims at the first order of blame. how dare you be so callous as to drop to an emotional attack as a discussion point. please look inwards at your sorry self.

so left hand fighting the right by that logic, either the education/infrastructure work wasn't/isn't working or the aid is too appealing...

no, I bring it up as a perfect example on how the w.h.o. came to a political solution despite masses of evidence to the contrary. every. single. time.

in terms of an age corrected population distribution, actually I think you'll find it's lower. that's the whole point of "at risk" in this scenario being mostly age >70yr (or wherever you chose to draw your risk factors) also this is tends to be over-estimating as the reported infection rates are failing to account for asymptomatic carriers which themselves follow an age-based distribution (I'm sorry but as far as a systematic goes, this is poorly recorded, poorly understood and very very poorly modelled, which is made even worse by the politicization of cv19 causing an mis-representation in people volunteering for randomized testing in a large number of western datasets). Again in very high population countries with poor vaccine rates, uptake and very slow roll-out we didn't end up with hundreds of thousands of bodies stacked high ready to be buried. that unfortunately is a large-scale uncomfortable truth to quite a few.

"no evidence of human to human transmission", was something repeated far too often and far too politically for me to take them serious on the next issue, serious or not.

> "It would spread when the carrier was <LARGELY> asymptomatic" , the largely is very important here otherwise containment would have been a lot different.

The main concerns for covid were also limited to a novel strain of a known virus type (again a KNOWN TYPE) being released into a global general populous with no inherent immunity. Aka expect ~5% of cases to probably have complications and some smaller %-age of that to be serious. If we didn't know what covid was we wouldn't be calling it "covid-19" to expressly describe which genus we're talking about. (Followed by general stupidity from people of pretending we don't know how other covid strains progress (regardless of any 'novel' effects)). Sill no sensible scenario put death rates >1% for anyone not in an at risk group. I mean everyone forgets the south-park sars skit that there's a 97% chance of catching that practically without symptoms. Why this became polarised about steam rolling through untested technology onto the populous is identical to the "green coal" and "tech will solve the carbon footprint" thinking...

Speaking from the academic sector if they're all able to meet ALL of the admissions criteria there would be no justification their presence, they would be in demand.

The sad reality is given the unrealistic expansion of the education sector they were clearly admitting people who needed to justify being there...

regardless of a reproduction science is repetition and verification yes... is there a point beyond that?

> Yes the movie was vaguely inspired by the science.

I.. _sigh_, my comment was satire I'm well aware of the "science", the modelling, the failures of large amounts of models, the ones that work, the...

Wait I've seen this... Day before the day after tomorrow right?

Good.

Lets hope we see less policy which is at a very small step back basically: "we're competing to punch ourselves in the face the hardest" in the international arena.

> we should send more money to climate scientists.

Couldn't disagree more.

Please spend it on those who might actually fix something. There's plenty of can remove carbon or can undo the effect of X on Y. Let's stop falling back on the bad argument of we must leave nature alone right after arguing we change billion dollar industries because we can.

We shouldn't learn to be custodians watching the planet die because of past mistakes, we should be fixing and improving the planet and improving on nature because we can, must and should, shoulder this reaponsibility.

Please not _yet more modelling_ burning HPC into the ground just for a crappy bar line graph (based on assumptions)...

> Whether it was fraudulent or just incorrect is a different question.

Sorry but good science should be reproducible an given the amount of major retractions around this field I think this starts to become more than a passing question.

Lots of work is built on top of the model which means a lot of money change hands and a lot of honesty hard work was done.

Those who built the bad model should be being asked why a) they were so wrong (bad conclusions are fine) b) their data was so bad (this is not defensible) c) they didn't push for reproducing their work sooner to support the field (more of a daming statement of a field which treats a statistical analysis as a push button tool)...

You. Wow.... You clearly aren't paying attention to how that movie was widely mid portrayed and presented within the education sector as fact

No I'm talking about proper healthy science not blind trust. Please don't confuse discussion with argument it's disingenuous and best I can say is look inwards.

Please don't bring up another thing started by idiot scientists for a laugh to laugh at stupid people. You have no idea what it's like dealing with the "just open your eyes" and "what else are they hiding" tier of pseudo-intellectualism enabled by nu-media.

There are reasons to be sceptical which are set in reason and it's worth not throwing that out with the bath water. Even if the bath water is full of low iq bitchute comments...

And believing the world ending as in "the day after tomorrow" was the "still mask wearing" of the 2010s. Fear.

I'll give you 2 reasons.

a) published data tends to see corrections from sensors and methodology which take several years to work out the fine details. (This isn't an attack this is science) Which means always take yesterday's numbers with more scepticism than 2yr ago. (This is making no statement of any data you're looking at or any trend you claim to see)

b) a field dominated by modelling needs data to back it up, otherwise the conversation would be, "Why is the LHC failing to find strong theory which is absolutely there" vs "I wonder if the modelling is correct based on..." This is a certain level of maturity that certain sciences are only starting to reach after playing in the ballpark of "let's go model my idea and make a press release which will just so happen to help my funding".

Yes sea level temps are rising, absolute numbers are still difficult to come by though and last UN summary doc I read still put things at 5C global average over a century. (Yes still horrifically catastrophic for the wrong people, but I'm also not in charge)

Serious question. Why are there static (in absolute positional terms) anomalies in the data that seem to be recording at the other end of the spectrum to their immediate surrounding waters?

Also nice to see several shipping lanes crop up when watching it.

I know many people from the continent who sound American because they learned ENGLISH language that way... yes it's strange how the world of communication centres around the world of discourse...

This isn't new. But nice to see more social sciences joining the party on the LLM bandwagon.

In defense of London from London...

Frankly if that's not an example of Stockholm syndrome I don't know what is.

It's so far from the nicest city on the country. It's entirely held up by "drip down" wealth from the banks and tourism which to paraphrase Chinese airways "avoid certain areas".

Flight from London is happening. It's already happened in a large way as well.

You don't see cockney anything anywhere there anymore compared to (and as much as you see any transitional regional identify left in) other British cities.

Homicide is on the drop in London but that's not 100% because it's safer. A huge amount is focused on deaths rather than attacks so don't fool yourself that just because they didn't die that nothing happened.

> crime which is a non-issue in the UK.

Nope. Not even close to true. Yes we don't have school shooters. Yes we don't have people exacting "justice" with a loaded barrel. But we do have gun crime and guns are used a LOT as intimidation. I wish I didn't grow up in an area where I know that to be true.

Trying to pretend there's not a problem is wonderful. And in that case I can point you to some very reasonably priced areas which must be perfectly safe and have no social cohesion issues at all regardless of where you're from...

Frankly given the sorry mid/long-term state of the UK economy, yes, this is the absolute essential thing to be moaning about...

And frankly moaning about this used to be right wing conspiracies a few years ago so yey for another pendulum swing...

So.

Is.

Life.

You've discovered probability, there was an 80% change of that. Roll a dice and do not pass go.

Again. The output from llm is a probable solution, not right, not wrong.

No. You should move students around more imo.

I've worked with good and bad at both. Some of the most difficult problems when you have students who have had excellent teachers and then get dropped into the real world. If they don't learn themselves how to apply what they're learning (the other side of the coin of training) then they're often no better than an llm stuck in a loop, they know the textbook but don't know the gray areas...

Also professors and researchers are required to be able to communicate otherwise they're useless to the field. They need to better.

I'm not saying every lecturer will hold any interest in every lecture course. I've had the ones who are there lecturing core material to avoid the dept losing its accreditation and I've done electives where the professor is off the wall and spends half of the time going on about their research instead of the address material (fun but painful come exam time).

You've clearly never worked the academic sector. Calm down most researchers are hyper focused on their research productivity because at times 90%+ of their time is consumed by teaching for months at a time. This is an almost universal constant for all decent institutes globally. Taking on extra cheap labor in the form of grad students used to be the only way to do this but every single time this turns into onboarding someone for months to get weeks of work out of them. Great when you can hide it in your other side of your job, but most of the time you can't...

very awesome writeup, glad to see someone with access to hw actually playing with this.

Hopefully the cost per GPU will kick-it soon and we'll see people properly play, but frankly the "middle section" layers 2(ish) to (n-1)(ish) of a model can be shuffled up/down and left/right and still perform well.

The fun one will be an LLM router for LLM layers to apply the best reasoning to the best input so far, but frankly that would need the years and years of training that the author hints at.

The one that's still out of grasps is still how to combine/manipulate per-layer k,v caches into a globally coherent state. i.e. if layers can be moved up/down why can't the cached k,v be swapped/combined with different projections? global k,v caches work, but they have to be _huge_ in order to prevent model collapse even on something as simple as owt.

> It’s sad to see Ars Technica at this level.

This was from a journalist _who_is_hired_as_expert_ at knowing of/about tooling that hallucinates (LLM ((AI)) chatbots). Decides to implicitly trust said technology to write a "hit piece" (lets be honest it was).

In several territories that would fall under slander and if is untrue is a major journalistic mis-step and career ending faux-pas.

Why in any situation would their position now be defendable?

This is akin to being a journalist of iron-mongering writing a "truth" piece on how "jet fuel can't melt steel beams" (if you don't get my reference here, lucky you). It's outright un-professional.

Blaming it on illness allows everyone to save face, but they were compos mentis enough to hit publish at the time. That itself carries a certain "I'm well enough to agree this is a good article" from said author.