🧛 Carnotaurus 🎃
banner
carnotaurus.bsky.social
🧛 Carnotaurus 🎃
@carnotaurus.bsky.social
Bluesky’s 4th most popular abelisaurid. Mostly doctor who / tech / paleo / general shitposts. Bad language and adult themes.

He/him/that
Unbelievable, this sucks.

If you’re feeling angry about those idiots, remember not to take it out on people who are on your side.
November 6, 2024 at 7:20 AM
Reposted by 🧛 Carnotaurus 🎃
My deep dive into the new Netflix #truecrime series, This is the Zodiac Speaking, are live mere hours before everything goes to hand in a hellbasket tomorrow. Lots of new, compelling evidence linking top suspect Arthur Leigh Allen in the series, but is it enough to give us any closure?
Was that the Zodiac Killer Speaking?
A new Netflix documentary series finds even more compelling evidence that Vallejo pedophile Arthur Leigh Allen was the Zodiac Killer, but does it bring us closure?A police sketch of the Zodiac killer,...
www.murdersthatmadeus.com
November 4, 2024 at 11:07 PM
Deal. Sorry mine are a little jaded.

1. The recent Dem optimism is going to turn out to be unfounded - the Selzer poll was just a bad one.

2. Harris will turn out to be less ahead among women than expected, and men will close the gap.
i think every political poster on Bluesky should be required to pre-register at least two election takes
November 4, 2024 at 11:11 PM
Still not over how Arriguccio finished #TheDecameron. What a superb show.
November 4, 2024 at 9:15 PM
Oh shit that thing is tomorrow? Guys don’t forget
November 4, 2024 at 6:30 PM
You know what the world really needs? A high level charting library for JavaScript. One that reinvents the grammar of graphics, except makes it really shit and incoherent. I don’t think we have one of those yet.
November 2, 2024 at 11:16 PM
Getting a bit tired of pearl-clutching over false or exaggerated ideas of AI rather than the actual social and ethical concerns. Like this story about the PLA using Llama. Militaries using open source code (despite license terms) isn’t a new thing.

www.reuters.com/technology/a...
Exclusive: Chinese researchers develop AI model for military use on back of Meta's Llama
Top Chinese research institutions linked to the People's Liberation Army have used Meta's publicly available Llama model to develop an AI tool for potential military applications, according to three academic papers and analysts.
www.reuters.com
November 2, 2024 at 10:11 AM
Here’s hoping this has the opposite effect to what was intended, and the credulous morons don’t ‘waste their effort’ voting
Crazy this is just background noise!
October 31, 2024 at 6:54 PM
Me, reading the paper on arXiv after spending a week failing to make my own version work
October 31, 2024 at 1:27 PM
Reposted by 🧛 Carnotaurus 🎃
New blog post up, detailing some of my recent experimentation with policy generation for NPC AIs. #gamedev #ml
hiivelabs.com/blog/gamedev...
Questions and comments welcomed.
Machine Learning Experiments, Part 1
This post is about the first in a series of experiments to determine the utility of using an LLM to build behavior policies for NPCs. Specifically, I am sett...
hiivelabs.com
October 30, 2024 at 2:57 PM
The comments on this are quite interesting. Fax machines are old and rubbish, and symbolise the private sector’s underinvestment in rail services. But also, if they still work then it makes sense to invest in improving other areas first.

Like the train operators have done with… errr… oh.
Extraordinary exchange captured by @jenwilliamsft.bsky.social on the other place - Andy Burnham challenged Northern Rail on their ongoing use of fax machines.
October 30, 2024 at 1:12 PM
Reposted by 🧛 Carnotaurus 🎃
Men, I challenge you to do this:

- No Sonic Screwdriver
- No army
- No plan
- 1-3 attractive young human or human like friends (one can be robot)
- 10,000 Daleks
- 1 TARDIS
- A Jammy Dodger
- Some jelly babies
- A yoyo
- Some string
- Save universe

Shall we?
October 30, 2024 at 6:42 AM
If I didn’t have BlueSky my work messages would be a lot less constructive, to say the least. Thank you, Internet friends
October 29, 2024 at 6:51 PM
A philosophy to live by, but especially for developers: don’t tear down a fence if you don’t understand why it was built in the first place.

theknowledge.io/chestertons-...
Chesterton's Fence: Explained
TL;DR / Summary G.K. Chesterton was an early 20th century English writer known for his clever paradoxes. He once wrote: “There exists in such a case a certain institution or law; let us say, for the...
theknowledge.io
October 29, 2024 at 6:47 PM
Devs: we don’t want to mindlessly copy your implementation, it’s important that we rethink it.

Me: yeah but instead you’re mindlessly building a worse implementation because you’re not thinking
Oh cool they also removed the one serious abstraction I put in place and just pissed the third party dependency all around the codebase.
Watching the devs convert my prototype to production code. Gleefully they stamp on every odd bit of code they see, not considering that maybe there’s a reason it’s like that. There is a reason, in the commit message. But it’s much more fun to re-introduce a bug and have to re-squash. What joy.
October 29, 2024 at 6:28 PM
Oh cool they also removed the one serious abstraction I put in place and just pissed the third party dependency all around the codebase.
Watching the devs convert my prototype to production code. Gleefully they stamp on every odd bit of code they see, not considering that maybe there’s a reason it’s like that. There is a reason, in the commit message. But it’s much more fun to re-introduce a bug and have to re-squash. What joy.
October 29, 2024 at 5:52 PM
Watching the devs convert my prototype to production code. Gleefully they stamp on every odd bit of code they see, not considering that maybe there’s a reason it’s like that. There is a reason, in the commit message. But it’s much more fun to re-introduce a bug and have to re-squash. What joy.
October 29, 2024 at 5:17 PM
Seems a super insightful and helpful thread, but once again: USA wtaf please don’t do this to people
It's open enrollment time so here's a convenient peg to hang a little tutorial about how to choose your health insurance.

Most of the advice you get is bad. Most of your instincts are bad. I however used to work for a health insurance company and I do arithmetic for a living so I can help.
Personally I think it's great that my company's HR sends out an official looking tool where they even ask you to pre-populate your healthcare providers and instead of being the actual insurance enrollment process it's just some bullshit checklist to try and convince me I actually don't want a PPO
October 28, 2024 at 10:58 PM
Reading up on the Mexico GP, apparently Verstappen got the penalty not because the stewards woke up, but because Norris kept ahead at the apex this time. Good for him for adapting. New racing guidelines due for Qatar that hopefully end the shenanigans once and for all. #F1
October 28, 2024 at 8:32 AM
On the one hand, I know better than to hope for Norris clinching an against-the-odds title victory. He needs to be P1 and Max P4 for every remaining race.

On the other hand, both Ferrari and McLaren are looking sharper than Red Bull, which would be the top 4 spots… #F1
October 26, 2024 at 7:47 PM
The FIA’s refusal to apply common sense on Verstappen’s moves is hugely frustrating and can’t be excused as anything other than bias. But I’m disappointed in the other #F1 drivers for not taking a stand too. Only Lewis has adapted his style, everyone else seems to just accept the disadvantage.
October 26, 2024 at 5:51 PM
Fighting GitHub actions. Programming in YAML and not finding out for several minutes whether your changes worked. What a heavenly combination.
October 26, 2024 at 4:17 PM
There’s a significant contingent of journalists who are MUCH more interested in being journalists than they are in doing journalism.
There is nothing more indicative of the problem than the Washington Post thinks 'we're not endorsing anyone' is a story that matters, but that 'one of the candidates ganged up with Jeffery Epstein to sexually assault a woman' isn't.
October 25, 2024 at 5:54 PM
Idly considering betting on the wrong one winning the US election, to try and soften the blow if it all goes wrong
October 25, 2024 at 5:09 PM
It’s just as well that Peter Capaldi only did three series of Doctor Who. He’s so incredibly good in the role he could easily have ended up owning it like Tom Baker did. Bit gutted he didn’t do a multi-Doctor adventure though (not counting the one with First).
October 25, 2024 at 11:27 AM