Martyn Plummer

@martynplummer.bsky.social

570 followers 190 following 79 posts

Statistics. Computing. Cancer Epidemiology. Public health.

Videos

Physics 31%

Public Health 25%

Posts Replies Media Videos

Martyn Plummer @martynplummer.bsky.social · 4d

This remake of Kind Hearts and Coronets is a bit dark.

www.vanityfair.com/style/story/...

Andrew Mountbatten Windsor Isn’t a Prince Anymore, but He Could Still Become King

King Charles’s brother remains eighth in the line of succession even though he no longer has his royal titles and honors.

www.vanityfair.com

Reposted by Martyn Plummer, Frédéric Barraquand

tj mahr 🎃 @tjmahr.com · 12d

student used some AI-assisted code to compute ROC stats with weights and the AI hallucinated pROC::roc(…, weights = weights). (the function doesn’t have a weights argument.) and because the function has a … argument, the fake weights argument was ignored without any warning

Martyn Plummer @martynplummer.bsky.social · 24d

The No Kings protests in the UK today have been helpfully renamed “No Tyrants”, just to clarify that they don’t want to get rid of the actual king.

Martyn Plummer @martynplummer.bsky.social · Oct 1

I think it’s plausible that, under the right conditions, life is common, but intelligent life is rare (it took billions of years on Earth). Without fossil fuels a sustainable industrial revolution might be next to impossible so industrial civilisations are even more rare.

Martyn Plummer @martynplummer.bsky.social · Sep 27

“All that was once directly lived has become mere representation”
- Guy Debord, The Society of the Spectacle (1967)

Financial Times @financialtimes.com · Sep 27

AI can’t raise the dead, but it might do the next best thing on.ft.com/4gFXyus | opinion

AI can’t raise the dead, but it might do the next best thing

Rapid advances in generative AI and voice technology mean communicating with the deceased will no longer be sci-fi

on.ft.com

Reposted by Martyn Plummer

Financial Times @financialtimes.com · Sep 27

AI can’t raise the dead, but it might do the next best thing on.ft.com/4gFXyus | opinion

AI can’t raise the dead, but it might do the next best thing

Rapid advances in generative AI and voice technology mean communicating with the deceased will no longer be sci-fi

on.ft.com

Martyn Plummer @martynplummer.bsky.social · Sep 24

"The ideal subject of totalitarian rule is not the convinced Nazi or the dedicated communist, but people for whom the distinction between fact and fiction, true and false, no longer exists." - Hannah Arendt

LBC @lbc.co.uk · Sep 24

"We were told thalidomide was a safe drug and it wasn't..."

Nigel Farage says he has 'no idea' if Donald Trump is right about paracetamol being linked to autism.

Reposted by Will Jennings, Steve Peers, Andrew Scott , and 9 more Will Jennings, Steve Peers, Andrew Scott, Tim Bale, Anand Menon, Jonathan Hopkin, David Merrett, Holger Hestermeyer, Steven Fielding, Ian Loader, James Connelly, Martyn Plummer

LBC @lbc.co.uk · Sep 24

"We were told thalidomide was a safe drug and it wasn't..."

Nigel Farage says he has 'no idea' if Donald Trump is right about paracetamol being linked to autism.

Reposted by Martyn Plummer

Adam Johansen @adamjohansen.bsky.social · Sep 23

There are a couple of vacancies in Wollongong for Maths/Stats lecturers:

www.uow.edu.au/about/jobs/j...

www.uow.edu.au

Martyn Plummer @martynplummer.bsky.social · Sep 22

People have been predicting the end of the world for some time and they’ve all been wrong so far. When the world failed to end in 1844, the Millerite movement called it “The Great Disappointment“. I don’t suppose that sentiment was widely shared.

Martyn Plummer @martynplummer.bsky.social · Sep 15

Yes. Melanoma is not the most common cancer in Australia but the others appear later. So when you restrict the picture to under-50s you get high peaks in Australia but also N Europe. Data from the IARC Global Cancer Observatory gco.iarc.who.int/en

Reposted by Martyn Plummer

R-Ladies @rladies.org · Sep 12

R Developer Day at University of Warwick brought together R-Ladies from around the world 🌎.

Spaces like this remind us how powerful our global community is when we learn, collaborate and build together.

Thanks to Heather Turner for making this possible 💜

#RDevDay
#rladies

“Smiling group of 14 R-Ladies at R Dev Day, posing together in a meeting room with a whiteboard and chairs in the background.

Reposted by Dorothy Bishop, Steve Peers, Elizabeth Stokoe , and 2 more Dorothy Bishop, Steve Peers, Elizabeth Stokoe, Martyn Plummer, Guy Dewsbury

LBC @lbc.co.uk · Sep 7

"It got the debate going and that's why we put him on."

Reform’s Laila Cunningham explains why her party allowed a 'quack' to falsely claim that the Covid vaccines gave the King cancer.

Dr. Lucky Tran @luckytran.com · Sep 4

The “problem” with vaccines? They so effective at preventing deaths that they create generations of people that question whether disease was a problem in the first place because they have never experienced the horrors of a world without vaccines.

Martyn Plummer @martynplummer.bsky.social · Sep 4

At the RSS conference in Edinburgh where @firthstat.bsky.social is presenting his new tetraplots for displaying 4-way results in UK elections. Wooden tetrahedron by @warwickstats.bsky.social woodworker-in-residence Nick Tawn.

David Firth holding a wooden tetrahedron in front of a poster displaying his new tetraplots. The proportion of votes for each of four parties is a point inside the volume of a tetrahedron, which is then projected onto the nearest surface and the tetrahedron is unfolded to make a flat display.

Reposted by Will Jennings, Chris Hanretty, Philip Cowley , and 1 more Will Jennings, Chris Hanretty, Philip Cowley, Martyn Plummer

David Firth @firthstat.bsky.social · Aug 26

At next week's RSS conference I'll be presenting a poster (a first for me...it's about time!)

"Tetraplot displays of UK General Election results" shows how to graph GE 2024 vote shares across 4 parties in a useful way.

Full PDF poster at:
github.com/DavidFirth/t...

Image of a poster to be presented at the Royal Statistical Society conference in Edinburgh, September 2025. The main point is to generalize the idea of a ternary plot to handle 4-part compositions, by showing views through the triangular faces of a tetrahedron. The data shown are vote counts from England, Scotland and Wales in the UK General Election of July 2024.

Martyn Plummer @martynplummer.bsky.social · Aug 24

It can’t give you advice because it doesn’t know anything. It’s a synthetic text generator that responds with a statistically plausible continuation of the context window based on the training data. @infinitescream.bsky.social

Martyn Plummer @martynplummer.bsky.social · Aug 24

Kyle [secretary of state for science innovation and technology] has been a vocal champion of AI … he had asked ChatGPT for advice on a range of work-related questions, including why British businesses were not adopting AI and what podcasts he should appear on.

www.theguardian.com/politics/202...

Deal to get ChatGPT Plus for whole of UK discussed by Open AI boss and minister

Exclusive: Deal that could have cost £2bn was floated at meeting between technology secretary Peter Kyle and Sam Altman

www.theguardian.com

Reposted by Martyn Plummer

F. Javier Rubio @fjrubio.bsky.social · Aug 21

New paper, with P. Basak, A.R. Linero, and C. Maringe, accepted in JASA A&CS

"Understanding Inequalities in Cancer Survival Using Bayesian Machine Learning"

doi.org/10.1080/0162...

#inequalities #cancer #survival #Bayesian #MachineLearning @icon-lshtm.bsky.social @statisticsucl.bsky.social

Martyn Plummer @martynplummer.bsky.social · Aug 18

This explains why Peter Kyle wants the Turing to focus on defense and security research which a) belongs in the public sector and b) is in the national interest.

It's a narrow view of the national interest but it answers the existential question.

Martyn Plummer @martynplummer.bsky.social · Aug 18

The problem with creating a centralised research institute is that you have to answer the question "Why here? Why can't this be done elsewhere?" Caught between industry and the university sector, the Alan Turing Institute is having a hard time answering.

The Guardian @theguardian.com · Aug 18

‘Shut it down and start again’: staff disquiet as Alan Turing Institute faces identity crisis

Whistleblower warns UK’s top AI research body in danger of collapse due to threats over funding and new direction

www.theguardian.com

Reposted by Martyn Plummer

The Guardian @theguardian.com · Aug 18

‘Shut it down and start again’: staff disquiet as Alan Turing Institute faces identity crisis

Whistleblower warns UK’s top AI research body in danger of collapse due to threats over funding and new direction

www.theguardian.com

Martyn Plummer @martynplummer.bsky.social · Aug 7

Gesundheit

Reposted by Martyn Plummer

Michael Friendly @datavisfriendly.bsky.social · Aug 7

#rstats I just submitted a new version of the {heplots} pkg to CRAN and got the message below.

It's August, and the #CRAN maintainers deserve a break, but it seems to me they also deserve a vote of thanks for all the work they do in vetting updates and new packages in R.

Screenshot of a message from CRAN stating that "CRAN submissions will be offline from Aug 01, 2025 to Aug 18, 2025" for CRAN team vacation and maintenance work.

Martyn Plummer @martynplummer.bsky.social · Aug 1

If I wanted to do this, I would add NA tags as a separate attribute of the data frame, not try to augment the numeric vector class.

Martyn Plummer @martynplummer.bsky.social · Aug 1

I think that tagged NAs are useful for data management but not necessarily for computation. You could extend the data frame class to allow tagged NAs, preserve them during data manipulation operations, and allow selection and transformation based on the tag.

Martyn Plummer @martynplummer.bsky.social · Aug 1

To some extent, yes. Critics of big tech have identified a pervasive set of cultish beliefs driving the AI industry that they call the #TESCREAL bundle.

www.dair-institute.org/tescreal/

The TESCREAL Bundle

The Distributed AI Research Institute is a space for independent, community-rooted AI research, free from Big Tech’s pervasive influence.

www.dair-institute.org

Martyn Plummer @martynplummer.bsky.social · Jul 31

Even if you have hardware and compiler support for NaN payload propagation, you still have to implement the semantics of NA vs NaN vs finite value propagation in all low-level arithmetic operations. Extending these semantics to include tagged NAs is too much complexity, and not guaranteed to work.

Martyn Plummer @martynplummer.bsky.social · Jul 31

NA_real_ in #rstats is implemented on top of IEEE754 arithmetic as NaN with a particular payload (1954). This seemed like a good idea at the time. Unfortunately, there is no guarantee that NaN payloads are propagated. It happens to work - mostly - on the platforms R is mainly used on.

Reposted by Maarten van Smeden, Martyn Plummer

Maarten van Smeden @maartenvsmeden.bsky.social · Feb 4

Periodic reminder the world of data analysis cannot be meaningfully categorised into "machine learning" and "statistics". Two cultures with substantial overlap in the use of methods (e.g. logistic regression), analytical goals (e.g. causal inference) and history

jamanetwork.com/journals/jam...

Light upyour news

Sign in to Lightnews

Sign up to start reading

Connect Bluesky

Connect with Bluesky

Light up
your news