Brendan Halpin
@bthalpin.bsky.social
Refugee from ex-Twitter, mostly at @[email protected]
Why the practice of changing the clock doesn't make sense too far south (say <40°) or north (>60°). Too far south the natural change is too small to bother, too far north too big for 1hr to make much difference. NB: much above 60° you don't have sunrise at all for part of the year.
November 3, 2025 at 10:27 PM
Why the practice of changing the clock doesn't make sense too far south (say <40°) or north (>60°). Too far south the natural change is too small to bother, too far north too big for 1hr to make much difference. NB: much above 60° you don't have sunrise at all for part of the year.
This idle observation prompted me to investigate the data on Irish-medium schools, using the disorganised data at the Dept of Ed www.gov.ie/en/departmen...,
a set of incompatible spreadsheets. After a lot of munging I get the following graph: approximately doubled between 1990s and 2020s:
a set of incompatible spreadsheets. After a lot of munging I get the following graph: approximately doubled between 1990s and 2020s:
October 26, 2025 at 8:00 PM
This idle observation prompted me to investigate the data on Irish-medium schools, using the disorganised data at the Dept of Ed www.gov.ie/en/departmen...,
a set of incompatible spreadsheets. After a lot of munging I get the following graph: approximately doubled between 1990s and 2020s:
a set of incompatible spreadsheets. After a lot of munging I get the following graph: approximately doubled between 1990s and 2020s:
October 25, 2025 at 8:34 PM
I had fun with OpenStreetMap data and Julia yesterday, so I wrote some notes on how I did it (i.e., access OSM data, process it with Shapefile.jl and plot it with Plots.jl).
Details at brendanhalpin.net/blog/posts/o...
Details at brendanhalpin.net/blog/posts/o...
October 3, 2025 at 11:13 PM
I had fun with OpenStreetMap data and Julia yesterday, so I wrote some notes on how I did it (i.e., access OSM data, process it with Shapefile.jl and plot it with Plots.jl).
Details at brendanhalpin.net/blog/posts/o...
Details at brendanhalpin.net/blog/posts/o...
I haven't been paying attention, but apparently it's been a bit stormy today.
Line graph showing a count of electrical outages in Ireland over 2+ weeks, mostly under 5, peaking at over 450 just now. Based on powercheck.ie.
#SpeirGhorm
Line graph showing a count of electrical outages in Ireland over 2+ weeks, mostly under 5, peaking at over 450 just now. Based on powercheck.ie.
#SpeirGhorm
October 3, 2025 at 9:43 PM
I haven't been paying attention, but apparently it's been a bit stormy today.
Line graph showing a count of electrical outages in Ireland over 2+ weeks, mostly under 5, peaking at over 450 just now. Based on powercheck.ie.
#SpeirGhorm
Line graph showing a count of electrical outages in Ireland over 2+ weeks, mostly under 5, peaking at over 450 just now. Based on powercheck.ie.
#SpeirGhorm
A map of Ireland (north & south), drawn only w/ motorways, primary & trunk roads, from OSM data provided as shapefiles from download.geofabrik.de/europe/irela...
Decoded and drawn entirely with Julia (and unzip, admittedly) using Shapefile.jl and Plots.jl.
#OpenStreetMap #JuliaLang #Speirghorm
Decoded and drawn entirely with Julia (and unzip, admittedly) using Shapefile.jl and Plots.jl.
#OpenStreetMap #JuliaLang #Speirghorm
October 2, 2025 at 9:23 PM
A map of Ireland (north & south), drawn only w/ motorways, primary & trunk roads, from OSM data provided as shapefiles from download.geofabrik.de/europe/irela...
Decoded and drawn entirely with Julia (and unzip, admittedly) using Shapefile.jl and Plots.jl.
#OpenStreetMap #JuliaLang #Speirghorm
Decoded and drawn entirely with Julia (and unzip, admittedly) using Shapefile.jl and Plots.jl.
#OpenStreetMap #JuliaLang #Speirghorm
#Brightspace is fucking up my teaching materials by being too clever.
It interprets latex code. Brilliant.
It just can't not do it, so my regexp example of
r"\([0-9]+\)")
is corrupted. Anything bracketed by \( & \) gets interpreted, even in tags, if the HTML is created outside BS.
It interprets latex code. Brilliant.
It just can't not do it, so my regexp example of
r"\([0-9]+\)")
is corrupted. Anything bracketed by \( & \) gets interpreted, even in tags, if the HTML is created outside BS.
![Appears as r"[0-9]+" with [0-9]+ in latex style, Computer Modern Roman font, and the \( and \) markers completely missing](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:ivvxsddno7zheyve4fiu67np/bafkreiewguz2k5zyvtu57gedctjduc7fpztv5kfgytc4smdkqte4dofy4i@jpeg)
September 30, 2025 at 11:53 PM
#Brightspace is fucking up my teaching materials by being too clever.
It interprets latex code. Brilliant.
It just can't not do it, so my regexp example of
r"\([0-9]+\)")
is corrupted. Anything bracketed by \( & \) gets interpreted, even in tags, if the HTML is created outside BS.
It interprets latex code. Brilliant.
It just can't not do it, so my regexp example of
r"\([0-9]+\)")
is corrupted. Anything bracketed by \( & \) gets interpreted, even in tags, if the HTML is created outside BS.
I see Steen hangs around with Marxists #Speirghorm
September 16, 2025 at 12:47 AM
I see Steen hangs around with Marxists #Speirghorm
Trump, Melania, Epstein and Ghislaine Maxwell. Three versions of the same shot, seconds apart. What is Epstein doing with his right hand, and why does Melania react like that?
FWIW, in photo 2 Melania's eyes are wider than in any other photo I've seen.
FWIW, in photo 2 Melania's eyes are wider than in any other photo I've seen.
July 18, 2025 at 8:39 PM
Trump, Melania, Epstein and Ghislaine Maxwell. Three versions of the same shot, seconds apart. What is Epstein doing with his right hand, and why does Melania react like that?
FWIW, in photo 2 Melania's eyes are wider than in any other photo I've seen.
FWIW, in photo 2 Melania's eyes are wider than in any other photo I've seen.
They'd be quite OK with this:
June 25, 2025 at 9:32 PM
They'd be quite OK with this:
After more investigation I'm beginning to think the data generating processes are indistinguishable. Intuition says that with the urn model, the sum to the previous element should be a better predictor of ti than under model 2, but the coefs from logit of ti on the sum to i-1 are indistinguishable.
June 11, 2025 at 12:43 PM
After more investigation I'm beginning to think the data generating processes are indistinguishable. Intuition says that with the urn model, the sum to the previous element should be a better predictor of ti than under model 2, but the coefs from logit of ti on the sum to i-1 are indistinguishable.
Summing the sequences yields an almost perfectly even distribution, so I suspect (1).
June 10, 2025 at 10:19 AM
Summing the sequences yields an almost perfectly even distribution, so I suspect (1).
My new pedagogical philosophy
May 18, 2025 at 10:39 PM
My new pedagogical philosophy
I've made a machine on my home network externally visible. Not without putting fail2ban on it, of course, to protect against ssh brute force attacks.
I decided to track fail2ban's activity, specifically the number of currently banned hosts. Constant attacks, but what a spike on May 6!
I decided to track fail2ban's activity, specifically the number of currently banned hosts. Constant attacks, but what a spike on May 6!
May 14, 2025 at 9:00 AM
I've made a machine on my home network externally visible. Not without putting fail2ban on it, of course, to protect against ssh brute force attacks.
I decided to track fail2ban's activity, specifically the number of currently banned hosts. Constant attacks, but what a spike on May 6!
I decided to track fail2ban's activity, specifically the number of currently banned hosts. Constant attacks, but what a spike on May 6!
Maybe this explains the model Boeing 747 that's been appearing in photos from the Oval Office recently. "Have you viewed the President's wish list recently?"
abcnews.go.com/Politics/tru...
abcnews.go.com/Politics/tru...
May 11, 2025 at 7:21 PM
Maybe this explains the model Boeing 747 that's been appearing in photos from the Oval Office recently. "Have you viewed the President's wish list recently?"
abcnews.go.com/Politics/tru...
abcnews.go.com/Politics/tru...
It took me a while to figure out that the problem is the sample standard error. The accuracy of the SE falls much faster than that of the mean or the SD (SE=SD/sqrt(N)!!!).
So I was pleased to see the sentence below in Gosset's original Biometrika paper (www.jstor.org/stable/2331554).
So I was pleased to see the sentence below in Gosset's original Biometrika paper (www.jstor.org/stable/2331554).
March 19, 2025 at 5:40 PM
It took me a while to figure out that the problem is the sample standard error. The accuracy of the SE falls much faster than that of the mean or the SD (SE=SD/sqrt(N)!!!).
So I was pleased to see the sentence below in Gosset's original Biometrika paper (www.jstor.org/stable/2331554).
So I was pleased to see the sentence below in Gosset's original Biometrika paper (www.jstor.org/stable/2331554).
Whose tongue is up whose ass? Trump is in awe of Putin, just like he used to be in awe of Epstein.
March 14, 2025 at 8:52 PM
Whose tongue is up whose ass? Trump is in awe of Putin, just like he used to be in awe of Epstein.
See also Verdi's opera "Lidl"
February 20, 2025 at 12:32 PM
See also Verdi's opera "Lidl"
For info, here's a (very compressed) image of the same data for 30 countries. GB is never an outlier (neither exceptionally high nor low).
February 14, 2025 at 5:54 PM
For info, here's a (very compressed) image of the same data for 30 countries. GB is never an outlier (neither exceptionally high nor low).
A graphic I've shared before: my go-to image for how the salience of immigration is driven by the media & pols, not by reality
It uses European Social Survey data (2020-2022-ish, every 2 yrs) 6 repeated Qs about immigration, for the UK
Takeaway: high salience until Brexit, then drops dramatically
It uses European Social Survey data (2020-2022-ish, every 2 yrs) 6 repeated Qs about immigration, for the UK
Takeaway: high salience until Brexit, then drops dramatically
February 14, 2025 at 4:43 PM
A graphic I've shared before: my go-to image for how the salience of immigration is driven by the media & pols, not by reality
It uses European Social Survey data (2020-2022-ish, every 2 yrs) 6 repeated Qs about immigration, for the UK
Takeaway: high salience until Brexit, then drops dramatically
It uses European Social Survey data (2020-2022-ish, every 2 yrs) 6 repeated Qs about immigration, for the UK
Takeaway: high salience until Brexit, then drops dramatically