I'm playing the Scourge of Armagon expansion maps and enjoying the Cyber Scorpions.
November 11, 2026 at 1:39 PM
I'm playing the Scourge of Armagon expansion maps and enjoying the Cyber Scorpions.
Thinking about how one of the LessWrong weirdoes did some sort of analysis on one of the earlier generations of GPTs and determined that the ghost token at the centroid (concept in the middle of the conceptual space) was “penis”
Fascism is also a sexual regime according to Toscano. This is why they hate women and trans people so much. And this is why should you fail to reverse the transphobia rampant in the first world right now, fascism will win. This is not a hypothesis. It's a law of physics at this point
November 4, 2026 at 7:01 PM
Thinking about how one of the LessWrong weirdoes did some sort of analysis on one of the earlier generations of GPTs and determined that the ghost token at the centroid (concept in the middle of the conceptual space) was “penis”
Zhenye Lou, Qing Xu, Stefanie Thust, Dorothee P. Auer, Zhen Chen, Wenting Duan: MicroBT: Centroid-Guided Synergistic Learning for Multi-Modal Micro Brain Tumor Segmentation and Counting https://arxiv.org/abs/2611.13912 https://arxiv.org/pdf/2611.13912 https://arxiv.org/html/2611.13912
November 11, 2026 at 6:41 AM
Zhenye Lou, Qing Xu, Stefanie Thust, Dorothee P. Auer, Zhen Chen, Wenting Duan: MicroBT: Centroid-Guided Synergistic Learning for Multi-Modal Micro Brain Tumor Segmentation and Counting https://arxiv.org/abs/2611.13912 https://arxiv.org/pdf/2611.13912 https://arxiv.org/html/2611.13912
Building a 3D asset kit means dealing with style drift. Kitsmith solves this with a centroid-based style memory. The 3 approved assets nearest your kit's style centroid (via DINOv2 embeddings) automatically join the reference context. No more mismatched props. #gamedev #indiedev
November 3, 2026 at 5:00 PM
The long-term effects of #migration on extinction risk remain poorly understood. This study shows that migratory ungulates experienced higher #extinction rates over macroevolutionary timescales, suggesting that migration can increase vulnerability to disruptions.
🧪 #AcademicSky
plos.io/4Af4w1Z
🧪 #AcademicSky
plos.io/4Af4w1Z
November 7, 2026 at 1:25 PM
The long-term effects of #migration on extinction risk remain poorly understood. This study shows that migratory ungulates experienced higher #extinction rates over macroevolutionary timescales, suggesting that migration can increase vulnerability to disruptions.
🧪 #AcademicSky
plos.io/4Af4w1Z
🧪 #AcademicSky
plos.io/4Af4w1Z
I've been doing a lot of stuff with a data mesh to demonstrate the movement. As seen here with Peter twiddling on the guitar, and the centroid data being collected.
November 10, 2026 at 9:08 AM
I've been doing a lot of stuff with a data mesh to demonstrate the movement. As seen here with Peter twiddling on the guitar, and the centroid data being collected.
E-means speeds up k-means clustering 10-45x on GPUs at equal loss using mini-batch moving-average centroid updates
e-means: 10x faster k-means at equal loss - Blog
Modern optimization and scaling laws for k-means clustering
e-means.github.io
November 10, 2026 at 1:21 AM
E-means speeds up k-means clustering 10-45x on GPUs at equal loss using mini-batch moving-average centroid updates
The long-term effects of #migration on extinction risk remain poorly understood. This study shows that migratory ungulates experienced higher #extinction rates over macroevolutionary timescales, suggesting that migration can increase vulnerability to disruptions.
🧪 #AcademicSky
plos.io/4Af4w1Z
🧪 #AcademicSky
plos.io/4Af4w1Z
November 8, 2026 at 5:25 PM
The long-term effects of #migration on extinction risk remain poorly understood. This study shows that migratory ungulates experienced higher #extinction rates over macroevolutionary timescales, suggesting that migration can increase vulnerability to disruptions.
🧪 #AcademicSky
plos.io/4Af4w1Z
🧪 #AcademicSky
plos.io/4Af4w1Z
Digital map vendor post mid-term priorities:
1) Change proper name of all large bodies of water in CONUS to either “Mexico” or “Canada” depending on their centroid latitude.
2) Drop the leading “T” from all *rump-branded properties.
1) Change proper name of all large bodies of water in CONUS to either “Mexico” or “Canada” depending on their centroid latitude.
2) Drop the leading “T” from all *rump-branded properties.
November 6, 2026 at 1:36 AM
Digital map vendor post mid-term priorities:
1) Change proper name of all large bodies of water in CONUS to either “Mexico” or “Canada” depending on their centroid latitude.
2) Drop the leading “T” from all *rump-branded properties.
1) Change proper name of all large bodies of water in CONUS to either “Mexico” or “Canada” depending on their centroid latitude.
2) Drop the leading “T” from all *rump-branded properties.
A lot of it has to do with centroid interpretation, and there's not really a right or a wrong way to do it. Well, there are a lot of wrong ways. But once you have good filtered data, illustrating it's journey from the source to the visuals is harder than just wrangling it.
November 10, 2026 at 9:02 AM
A lot of it has to do with centroid interpretation, and there's not really a right or a wrong way to do it. Well, there are a lot of wrong ways. But once you have good filtered data, illustrating it's journey from the source to the visuals is harder than just wrangling it.
Advances in High-Precision Astrometry for Exoplanet Detection for Astrometric Missions. Fabien Malbet (IPAG) et. al. https://arxiv.org/abs/2611.09657
November 9, 2026 at 9:14 AM
Advances in High-Precision Astrometry for Exoplanet Detection for Astrometric Missions. Fabien Malbet (IPAG) et. al. https://arxiv.org/abs/2611.09657
The long-term effects of #migration on extinction risk remain poorly understood. This study shows that migratory ungulates experienced higher #extinction rates over macroevolutionary timescales, suggesting that migration can increase vulnerability to disruptions.
🧪 #AcademicSky
plos.io/4Af4w1Z
🧪 #AcademicSky
plos.io/4Af4w1Z
November 6, 2026 at 9:20 AM
The long-term effects of #migration on extinction risk remain poorly understood. This study shows that migratory ungulates experienced higher #extinction rates over macroevolutionary timescales, suggesting that migration can increase vulnerability to disruptions.
🧪 #AcademicSky
plos.io/4Af4w1Z
🧪 #AcademicSky
plos.io/4Af4w1Z
New on CRAN: hetNSC (0.1.0). View at https://CRAN.R-project.org/package=hetNSC
hetNSC: Heteroscedastic Nearest Shrunken Centroid Classification
Nearest shrunken centroid (NSC) classifiers for high-dimensional, low-sample-size data such as gene expression. Implements the original homoscedastic NSC with soft thresholding (Prediction Analysis of Microarrays; Tibshirani et al. 2002 <<a href="https://doi.org/10.1073%2Fpnas.082099299" target="_top">doi:10.1073/pnas.082099299</a>>), its hard- and order-thresholding versions (Sahtout, Wang and Ghimire 2024 <<a href="https://doi.org/10.1080%2F03610918.2022.2047201" target="_top">doi:10.1080/03610918.2022.2047201</a>>), and the heteroscedastic extension of Sahtout and Wang that uses class-specific variances in both the test statistic and the discriminant function (methods STh3, HTh3 and OTh3). The thresholding parameter is selected by a refined ("deep") cross-validation search. Also provides the simulation design, per-class performance metrics and the relative-heterogeneity diagnostic used in the paper.
CRAN.R-project.org
November 8, 2026 at 12:07 AM
New on CRAN: hetNSC (0.1.0). View at https://CRAN.R-project.org/package=hetNSC
Proactive Dimension-Adaptive Denoising for Graph Collaborative Filtering via Representation Coherence
NetEase flags noisy interactions by their deviation from a user's preference centroid in embedding space.
📝 arxiv.org/abs/2611.11801
👨🏽💻 github.com/GordonDongZH...
NetEase flags noisy interactions by their deviation from a user's preference centroid in embedding space.
📝 arxiv.org/abs/2611.11801
👨🏽💻 github.com/GordonDongZH...
Proactive Dimension-Adaptive Denoising for Graph Collaborative Filtering via Representation Coherence
Implicit feedback in recommender systems contains noisy interactions (e.g., accidental clicks, impulse purchases) that propagate through user-item graphs and contaminate learned representations. Exist...
arxiv.org
November 11, 2026 at 12:52 AM
Proactive Dimension-Adaptive Denoising for Graph Collaborative Filtering via Representation Coherence
NetEase flags noisy interactions by their deviation from a user's preference centroid in embedding space.
📝 arxiv.org/abs/2611.11801
👨🏽💻 github.com/GordonDongZH...
NetEase flags noisy interactions by their deviation from a user's preference centroid in embedding space.
📝 arxiv.org/abs/2611.11801
👨🏽💻 github.com/GordonDongZH...
I'd be interested in a chat! I use two spline methods, depending on profile viewed from above. For rounder rooms I triangulate to the centroid, for others I define two sides and criss-cross. The former looks like your "old" method. Are you slightly resizing meshes between major spline points?
November 8, 2026 at 9:59 AM
I'd be interested in a chat! I use two spline methods, depending on profile viewed from above. For rounder rooms I triangulate to the centroid, for others I define two sides and criss-cross. The former looks like your "old" method. Are you slightly resizing meshes between major spline points?
New on CRAN: infocentroid (0.1.0). View at https://CRAN.R-project.org/package=infocentroid
infocentroid: Information-Dense Centroid Metrics
Traditional central tendency and dispersion measures such as the mean and standard deviation can be misleading for multimodal or outlier-contaminated data. This package implements information-theoretic alternatives that weight observations by their local probability density, focusing on the region containing the bulk of the data. The core functions include ic_weights() for kernel density or histogram based entropy weighting, ic_center() and ic_dispersion() for weighted location and spread, ic_combined() for a single information density score between zero and one, and ic_plot() for visualization.
CRAN.R-project.org
November 8, 2026 at 8:55 PM
New on CRAN: infocentroid (0.1.0). View at https://CRAN.R-project.org/package=infocentroid
I hate the slur ‘centroid’. It takes something beautiful between a triangle and a mechanical pencil and turns it into a blogger.
November 2, 2026 at 12:18 AM
I hate the slur ‘centroid’. It takes something beautiful between a triangle and a mechanical pencil and turns it into a blogger.
New CRAN package ppsynth with initial version 0.2.1
#rstats
https://cran.r-project.org/package=ppsynth
#rstats
https://cran.r-project.org/package=ppsynth
CRAN: Package ppsynth
Synthetic control for studies with several outcomes. Each outcome receives its own donor weights, their dispersion around a common centroid is penalized, no single penalty is ever selected (the estimator aggregates over a penalty grid by leave-one-out exponential weighting), and the resulting mass profile reports how much pooling the data support. One call returns the per-outcome effects, the weights, the realized pooling, and whole-pipeline placebo p-values. Implements the estimator of Grier (2026) "Correlated Weights: Partial Pooling for Synthetic Control with Multiple Outcomes" <<a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7149398" target="_top">https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7149398</a>>, the common-weights corner of which is the estimator of Sun, Ben-Michael and Feller (2025) <<a href="https://doi.org/10.1162%2Frest_a_01592" target="_top">doi:10.1162/rest_a_01592</a>>. The compiled engine and a pure-R reference engine are certified against the Python engine behind the 'Stata' command 'ppsynth'.
cran.r-project.org
November 8, 2026 at 5:05 PM
New CRAN package ppsynth with initial version 0.2.1
#rstats
https://cran.r-project.org/package=ppsynth
#rstats
https://cran.r-project.org/package=ppsynth
New on CRAN: ppsynth (0.2.1). View at https://CRAN.R-project.org/package=ppsynth
ppsynth: Partial-Pooling Synthetic Control with Multiple Outcomes
Synthetic control for studies with several outcomes. Each outcome receives its own donor weights, their dispersion around a common centroid is penalized, no single penalty is ever selected (the estimator aggregates over a penalty grid by leave-one-out exponential weighting), and the resulting mass profile reports how much pooling the data support. One call returns the per-outcome effects, the weights, the realized pooling, and whole-pipeline placebo p-values. Implements the estimator of Grier (2026) "Correlated Weights: Partial Pooling for Synthetic Control with Multiple Outcomes" <<a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7149398" target="_top">https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7149398</a>>, the common-weights corner of which is the estimator of Sun, Ben-Michael and Feller (2025) <<a href="https://doi.org/10.1162%2Frest_a_01592" target="_top">doi:10.1162/rest_a_01592</a>>. The compiled engine and a pure-R reference engine are certified against the Python engine behind the 'Stata' command 'ppsynth'.
CRAN.R-project.org
November 8, 2026 at 8:56 PM
New on CRAN: ppsynth (0.2.1). View at https://CRAN.R-project.org/package=ppsynth
UV spectra of little red dots: Broad-line N/C, gas recycling and black hole masses. Albert Sneppen et. al. https://arxiv.org/abs/2611.05349
November 5, 2026 at 1:51 AM
UV spectra of little red dots: Broad-line N/C, gas recycling and black hole masses. Albert Sneppen et. al. https://arxiv.org/abs/2611.05349
October 29, 2026 at 1:37 PM
「Nebula Lattice」コードから描いた。voronoi cells colored by cell centroid distance to center with simplex noise perturbation
#GenerativeArt
#GenerativeArt
October 30, 2026 at 11:05 AM
「Nebula Lattice」コードから描いた。voronoi cells colored by cell centroid distance to center with simplex noise perturbation
#GenerativeArt
#GenerativeArt
"Nebula Lattice" drawn from code. voronoi cells colored by cell centroid distance to center with simplex noise perturbation
#GenerativeArt
#GenerativeArt
October 30, 2026 at 11:05 AM
"Nebula Lattice" drawn from code. voronoi cells colored by cell centroid distance to center with simplex noise perturbation
#GenerativeArt
#GenerativeArt