Surprise post! Randomisation errors and how to avoid sabotaging your model with bad use of randomisation! idlecartulary.com/2026/01/06/w... #osr
What to randomise when you’re randomising
You may have noticed that a common theme in my reviews is that I criticise people’s random tables: Their wilderness generators, their wandering monsters, their weather tables, their treasure …
idlecartulary.com
January 5, 2026 at 9:44 PM
Surprise post! Randomisation errors and how to avoid sabotaging your model with bad use of randomisation! idlecartulary.com/2026/01/06/w... #osr
Former QA Manager here.
If you put randomisation elements in your game I am going to haunt your ass
If you put randomisation elements in your game I am going to haunt your ass
March 29, 2026 at 7:04 AM
Former QA Manager here.
If you put randomisation elements in your game I am going to haunt your ass
If you put randomisation elements in your game I am going to haunt your ass
while the British Academy is using partial randomisation on its small grants -- and awarding more women/and POC researchers than ever before -- UKRI is feeding your painstakingly-developed proposal into an AI. But it's okay because they then get *two* AIs to debate whether you get the money.
Had a UKRI funded grant proposal rejected at the "AI triage stage". "This AI-based triage was used only to identify approximately 50% of the strongest proposals to take forward to full human review." 🧪 #academicsky
September 24, 2026 at 10:04 PM
while the British Academy is using partial randomisation on its small grants -- and awarding more women/and POC researchers than ever before -- UKRI is feeding your painstakingly-developed proposal into an AI. But it's okay because they then get *two* AIs to debate whether you get the money.
Some grant bodies have actually started to sift by lottery / randomisation to acknowlege the difficulty and difficulty of being truly objective: www.thebritishacademy.ac.uk/news/partial...
Partial randomisation trial extended after diversity of applicants and award holders increases
We're extending the partial randomisation trial on our Small Research Grants scheme for a further three years.
www.thebritishacademy.ac.uk
September 25, 2026 at 8:35 AM
Some grant bodies have actually started to sift by lottery / randomisation to acknowlege the difficulty and difficulty of being truly objective: www.thebritishacademy.ac.uk/news/partial...
But per-protocol analyses are difficult. Whether a doctor chooses to use the "AI stethoscope" is not determined at random.
This 'breaks' the randomisation, making the design equivalent to an observational study.
🧵 6/
This 'breaks' the randomisation, making the design equivalent to an observational study.
🧵 6/
February 25, 2026 at 4:44 PM
But per-protocol analyses are difficult. Whether a doctor chooses to use the "AI stethoscope" is not determined at random.
This 'breaks' the randomisation, making the design equivalent to an observational study.
🧵 6/
This 'breaks' the randomisation, making the design equivalent to an observational study.
🧵 6/
10) Exponential Backoff (with randomisation) - a great way to avoid accidentally DDoSing things with retry logic
October 24, 2024 at 1:47 AM
10) Exponential Backoff (with randomisation) - a great way to avoid accidentally DDoSing things with retry logic
💷🇬🇧Has randomisation made the British Academy's (@britishacademy.bsky.social) grants more attractive by ensuring equal chances and removing bias?
Read here 👉 ow.ly/ZrAt50VIJSK
#Research #Grants #Funding #Randomisation
Read here 👉 ow.ly/ZrAt50VIJSK
#Research #Grants #Funding #Randomisation
April 28, 2025 at 10:18 AM
💷🇬🇧Has randomisation made the British Academy's (@britishacademy.bsky.social) grants more attractive by ensuring equal chances and removing bias?
Read here 👉 ow.ly/ZrAt50VIJSK
#Research #Grants #Funding #Randomisation
Read here 👉 ow.ly/ZrAt50VIJSK
#Research #Grants #Funding #Randomisation
On my way to London from Edinburgh to make sure I can attend the discussion paper on randomisation tomorrow.
rss.org.uk/training-eve...
rss.org.uk/training-eve...
‘Randomisation in Clinical Trials: Is a Pragmatic Compromise between Randomisation- and Model-Based
rss.org.uk
June 30, 2026 at 7:17 PM
On my way to London from Edinburgh to make sure I can attend the discussion paper on randomisation tomorrow.
rss.org.uk/training-eve...
rss.org.uk/training-eve...
British Academy randomisation trial boosts award diversity.
Proportion of awards granted to Black and Asian applicants increases.
www.researchprofessionalnews.com/rr-news-uk-c...
Proportion of awards granted to Black and Asian applicants increases.
www.researchprofessionalnews.com/rr-news-uk-c...
April 16, 2025 at 6:09 AM
British Academy randomisation trial boosts award diversity.
Proportion of awards granted to Black and Asian applicants increases.
www.researchprofessionalnews.com/rr-news-uk-c...
Proportion of awards granted to Black and Asian applicants increases.
www.researchprofessionalnews.com/rr-news-uk-c...
2/ Emphasising the role of randomisation. Randomisation is entirely compatible with the use of routinely collected data. Importantly, the source of the data does not compensate for weaknesses in study design, and "real-world evidence" is no substitute for rigorous methodology...
June 2, 2026 at 1:54 PM
2/ Emphasising the role of randomisation. Randomisation is entirely compatible with the use of routinely collected data. Importantly, the source of the data does not compensate for weaknesses in study design, and "real-world evidence" is no substitute for rigorous methodology...
For the solutions, I highlighted that lotteries actually have more potential with grants and fellowships than things like college admissions. In a 3 year pilot of partial randomization, the UK funded more racially and institutionally diverse projects. 3/
www.thebritishacademy.ac.uk/news/partial...
www.thebritishacademy.ac.uk/news/partial...
Partial randomisation trial extended after diversity of applicants and award holders increases
We're extending the partial randomisation trial on our Small Research Grants scheme for a further three years.
www.thebritishacademy.ac.uk
April 17, 2026 at 12:57 PM
For the solutions, I highlighted that lotteries actually have more potential with grants and fellowships than things like college admissions. In a 3 year pilot of partial randomization, the UK funded more racially and institutionally diverse projects. 3/
www.thebritishacademy.ac.uk/news/partial...
www.thebritishacademy.ac.uk/news/partial...
Loads of evidence from e.g. UK and Europe that partial randomisation can proactively enhance diversity of applicant pool and awardee pool. What all responsible funders need to test too is what (if any) impact this allocation method has on outcomes. (To date, know of no reported negative impacts).
May 13, 2026 at 8:49 AM
Loads of evidence from e.g. UK and Europe that partial randomisation can proactively enhance diversity of applicant pool and awardee pool. What all responsible funders need to test too is what (if any) impact this allocation method has on outcomes. (To date, know of no reported negative impacts).
Some musing on Mendelian Randomisation as a technique, triggered by discussions with a variety of colleagues. TL;DR Mendelian Randomisation works well when it used for hypothesis testing of valid exposure => outcome scenarios, as long as the MR assumptions hold (obvs!) and it is performed with care.
July 26, 2025 at 6:30 AM
Some musing on Mendelian Randomisation as a technique, triggered by discussions with a variety of colleagues. TL;DR Mendelian Randomisation works well when it used for hypothesis testing of valid exposure => outcome scenarios, as long as the MR assumptions hold (obvs!) and it is performed with care.
Agree. Stephen Senn's "Seven myths of randomisation in clinical trials" pubmed.ncbi.nlm.nih.gov/23255195/ is a good place to start.
And the work by Jamie Robins and colleagues helped us understand "the curse of dimensionality" in high-dimensional settings (references in Chapter 10 of "What If").
And the work by Jamie Robins and colleagues helped us understand "the curse of dimensionality" in high-dimensional settings (references in Chapter 10 of "What If").
Seven myths of randomisation in clinical trials - PubMed
I consider seven misunderstandings that may be encountered about the nature, purpose and properties of randomisation in clinical trials. Some concern the practical realities of clinical research on pa...
pubmed.ncbi.nlm.nih.gov
November 26, 2024 at 1:56 PM
Agree. Stephen Senn's "Seven myths of randomisation in clinical trials" pubmed.ncbi.nlm.nih.gov/23255195/ is a good place to start.
And the work by Jamie Robins and colleagues helped us understand "the curse of dimensionality" in high-dimensional settings (references in Chapter 10 of "What If").
And the work by Jamie Robins and colleagues helped us understand "the curse of dimensionality" in high-dimensional settings (references in Chapter 10 of "What If").
With large enough n you don't need randomisation.
July 1, 2026 at 8:52 PM
With large enough n you don't need randomisation.
Check out our newest InsideGMT Article: "Randomisation and emergent narratives in Combat Commander: Vietnam" by Joe Dewhurst. zurl.co/jsjEG
August 12, 2026 at 2:19 AM
Check out our newest InsideGMT Article: "Randomisation and emergent narratives in Combat Commander: Vietnam" by Joe Dewhurst. zurl.co/jsjEG
Today I’m remembering the time my twin brother and I hit the jackpot buying a chocolate paddle-pop from the van-cafe at the park and for some reason the ice cream factory had borked the randomisation of the prize-sticks and every stick won us a free paddle-pop
January 30, 2026 at 6:44 AM
Today I’m remembering the time my twin brother and I hit the jackpot buying a chocolate paddle-pop from the van-cafe at the park and for some reason the ice cream factory had borked the randomisation of the prize-sticks and every stick won us a free paddle-pop
The winner by randomisation is... @justgreenie.bsky.social 🥳 CONGRATS 💛✨
1000th Post to Bsky and my ROAD TO 1K RAFFLE!!! 💛✨
Ends at 1000 Followers.
To Enter:
- Follow me
- Comment any Emoji
Good luck!!! 🥳
Ends at 1000 Followers.
To Enter:
- Follow me
- Comment any Emoji
Good luck!!! 🥳
March 29, 2025 at 3:07 PM
The winner by randomisation is... @justgreenie.bsky.social 🥳 CONGRATS 💛✨
Organised my lingerie stash so I can come up with some sorta randomisation system when I post.
June 3, 2025 at 4:32 PM
Organised my lingerie stash so I can come up with some sorta randomisation system when I post.
The British Academy is extending partial randomisation of its Small Research Grants awards process to 2028. The first three years of the trial showed increased diversity of applicants, awardees, and universities. buff.ly/FlwQuEn
Partial randomisation trial extended after diversity of applicants and award holders increases
We're extending the partial randomisation trial on our Small Research Grants scheme for a further three years.
www.thebritishacademy.ac.uk
April 16, 2025 at 9:30 AM
The British Academy is extending partial randomisation of its Small Research Grants awards process to 2028. The first three years of the trial showed increased diversity of applicants, awardees, and universities. buff.ly/FlwQuEn
New paper!
‘Categorisation of continuous covariates for stratified randomisation: How should we adjust?’
Led by Tom Sullivan, featuring Brennan Kahan, Alana Cuthbert and Lisa Yelland.
@brennankahan.bsky.social
onlinelibrary.wiley.com/doi/10.1002/...
‘Categorisation of continuous covariates for stratified randomisation: How should we adjust?’
Led by Tom Sullivan, featuring Brennan Kahan, Alana Cuthbert and Lisa Yelland.
@brennankahan.bsky.social
onlinelibrary.wiley.com/doi/10.1002/...
March 19, 2024 at 11:30 AM
New paper!
‘Categorisation of continuous covariates for stratified randomisation: How should we adjust?’
Led by Tom Sullivan, featuring Brennan Kahan, Alana Cuthbert and Lisa Yelland.
@brennankahan.bsky.social
onlinelibrary.wiley.com/doi/10.1002/...
‘Categorisation of continuous covariates for stratified randomisation: How should we adjust?’
Led by Tom Sullivan, featuring Brennan Kahan, Alana Cuthbert and Lisa Yelland.
@brennankahan.bsky.social
onlinelibrary.wiley.com/doi/10.1002/...
Make Generic Drums Sound Insane : Mining MIDI for randomisation using 2 techniques
Mining MIDI for randomisation using 2 techniques
Make Generic Drums Sound Insane : Mining MIDI for randomisation using 2 techniques
sonicstate.com
July 15, 2026 at 9:44 AM
Make Generic Drums Sound Insane : Mining MIDI for randomisation using 2 techniques
Had to finish this thread in a rush without a concluding flourish but adding it here for my “how to do Mendelian Randomisation without messing up”
Some musing on Mendelian Randomisation as a technique, triggered by discussions with a variety of colleagues. TL;DR Mendelian Randomisation works well when it used for hypothesis testing of valid exposure => outcome scenarios, as long as the MR assumptions hold (obvs!) and it is performed with care.
July 26, 2025 at 6:47 AM
Had to finish this thread in a rush without a concluding flourish but adding it here for my “how to do Mendelian Randomisation without messing up”
💥New: More funders are turning to randomisation to allocate grants fairly when competition is fierce. Ken Emond, the Head of Research Funding at the British Academy (@britishacademy.bsky.social) shares how they're using it for Small #Research #Grants, & the positive outcomes after 3 years.
How randomisation has changed the British Academy’s approach to research funding - Impact of Social Sciences
As more research funders adopt randomisation, the British Academy explains why it has used partial randomisation to allocate its Small Research Grants scheme.
blogs.lse.ac.uk
April 16, 2025 at 11:21 AM
💥New: More funders are turning to randomisation to allocate grants fairly when competition is fierce. Ken Emond, the Head of Research Funding at the British Academy (@britishacademy.bsky.social) shares how they're using it for Small #Research #Grants, & the positive outcomes after 3 years.