LightNews LightNews
dodiscimus.bsky.social
Matt Perks
@dodiscimus.bsky.social
1.1K followers 2.4K following 1.8K posts
11-18 physics teacher, now at University of Southampton in Initial Teacher Education. School governor. Mostly edu-guessing. By teaching, we learn!
Posts Media Videos Starter Packs
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
Maybe it is in Ireland???

Generally, though, attention on international comparisons has waned since the heyday of looking to Finland and Singapore.

But, yes, worth asking what Ireland might be doing between Y5 and Y10. But it could also be cultural.
2 1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
Yes.
But El Niño is the reason the year is setting records, I think.
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
This is well worth reading 👇
A tiny bit of extrapolation around level of guidance, I think, but 98% spot on (and much better than I could manage).
1 1 1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
... we find a technology to sequester C at scale, the next generations will either have to adapt to very high levels of impact or attempt geo-engineering.
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
Some of this is El Niño but there's no doubt CO2 is still going up, and still going up fast.
Decarbonisation, renenwable electricity generation, etc. is making it less bad but I think +1.5°C is long gone, +2.0°C via net zero is a fantasy, and unless...
sustainable2050.bsky.social Kees van der Leun @sustainable2050.bsky.social · 19h
This graph says it all: instead of starting to go down, the *annual increase* of atmospheric CO2 is setting new records. We're making climate change worse at a record rate.
Turning this around should be our top priority, and it's not.
From: www.carbonbrief.org/met-office-a...
2
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
The other thing that tends to get lost under the rankings, is the rich data about well-being, and enjoyment of / participation in reading for pleasure.
If I was SoS, I'd be looking hard at that.
1 2
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
And I don't know how much PIRLS and PISA are affected by things like Y5 being keener to do well, or better supported by teachers in relation to the tests, or the nature of the test suiting them better than Y10/11 (compared to children in other countries).
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 7h
The published rankings for PISA are for UK whereas PIRLS are for England. The confidence intervals mean that there is no statistically significant difference between similar rankings. For example, there are only 8 education systems with scores significantly above England in PISA 2022 (not 13).
1 1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 8h
That is certainly reasonable speculation. I think more sustained reading (by and to) children at secondary school, particularly in subjects other than English, would be a good move.
OTOH I'm not sure the gap between PISA and PIRLS is as big as it looks (an expert opinion would be useful here).
1 1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 18h
Pianist (one of the music teachers from work).
Was a great decision.
Been to too many weddings where 6 kids and 2 drunk 50-somethings were the only people dancing and everyone else was having to yell at each other to make themselves heard.
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 18h
I think it would be fair to say English children generally do well on international comparison tests for reading, and have improved in the last ~decade relative to other countries.

Individual test series and rankings fluctuate.
1 1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
Not sure I can fully unpick the LLM v human comparison, but clear that human evaluation is very poor compared to Newsguard - at the flipping a coin level, if I'm reading that right.
Whereas LLMs are in much closer agreement with Newsguard, if a bit more sceptical.
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
LLM models consistently underestimated the reliability of right-leaning sources and the opposite with left-leaning ones.
Think it's speculative, but suggestion is that this is a function of the amount of association in the training data between right-leaning sources and criticism of reliability.
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
I'm not sure if this is sarcasm.

For me, a lot of that part of the paper seems like academic navel-gazing. The number of papers I've read that have felt the need to introduce a new word, I'm thinking it's a good job no-one relies on printed dictionaries these days. But the results are interesting.
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
Yes, nothing like "good-natured" violence to bring a community together and make everyone want to go to the footie with their kids.
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
Interesting paper here academic.oup.com/jeea/advance...
Looks like full marks to Chile for genuine evidence-based decision-making.
Seems to indicate a substantial downside to an attempt to level the university-admission playing field for low-SES students.
The Persistent Effect of Competition on Prosociality
Abstract. We present the first causal evidence on the persistent impact of enduring competition on prosociality. Inspired by the literature on tournaments
academic.oup.com
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
This failure is obviously (yet another) indication that AI adoption in education is not looking good.
However,
In extrapolating to a UK context, we need to remember,
(a) We don't have mandatory textbooks at the heart of our curriculum.
(b) This is a quality control failure, really.
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
This is a man making a big thing about anti-social behaviour on the tube...
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
Two things I don't miss about the 1980s:
1. The unrelenting procession of murders in (or in relation to) NI.
2. A culture in football where people regularly kicked the 💩 out of each other and smashed things up.
1 4
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
I do wonder how many Green voters might see what they've got here and go LibDem next time.
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 1d
I remember the Shameless Opportunism v Millwall games from the days when if you were in the wrong place at the wrong time you were likely to get your head kicked in.
GBH is just banter, really, for proper football fans.
1 4
Reposted by Matt Perks
anildash.com
Anil Dash @anildash.com · 2d
"I have a decent fluency in LLMs, and they have utility, but the absurd degree of over-hype, the way they're being forced on everyone, and the insistence on ignoring the many valid critiques about them make it very difficult to focus on legitimate uses where they might add value."
29 230 1.4K
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 2d
🤮
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 2d
TBF, my parents are in their 80s and I'm pretty sure they're not having children any time soon.

You know what, though. I'm used to senior members of, say, the Russian or Iranian government saying things which are patently untrue. It's hard when it's cumming from the USA.
1
dodiscimus.bsky.social
Matt Perks @dodiscimus.bsky.social · 2d
I think it's more likely to do the opposite, but if it did boost learning, would knowing that not be a substantial part of answering the "is it good?" question?