I never knew a possum was a marsupial.
Live and learn (but only if you choose too).
Judging by YouTube there are quite a number of people who take in young animals who’ve lost their mother and would die without help.
They feed and raise these little creatures until they are mature enough to be released back into the wild. Sometimes they leave, sometimes they keep coming back.
Here’s a young woman who took a baby opossum in. They bonded so thoroughly that the opossum would always come back. She has accepted that they will be together.
It’s believed that marsupials originally evolved in South America somewhere between 125–160 million years ago and migrated to Australia through (a then unglaciated) Antarctica via the Antarctic land bridge that existed for awhile during the Cretaceous period. They really couldn’t compete with placentals too well, so they only persisted in South America and Australia, both of which remained isloated from the other continents until South America came into contact with North America sometime around 15 million years ago.
Yes I know the physical difference between a o’ possum and a possum, still they’re all possums to me.
I didn’t even know that there were possums in Ireland!
And they are immune to rattlesnake venom.
Madison Kanna wrote at X: ‘as a software engineer, i feel a real loss of identity right now.
for a long time i defined myself in part by the act of writing code. the pride in a hard-earned solution was part of who i was. now i watch AI accomplish in seconds what took me hours. i find myself caught between relief and mourning, awe and anxiety. the craft that shaped me is suddenly eclipsed by a machine. who am i now?’
Her post inspired a post & discussion about skill obsolescence in historical perspective:
Madison Kanna wrote at X: ‘as a software engineer, i feel a real loss of identity right now….i watch AI accomplish in seconds what took me hours….
There are plenty of software engineers who don’t think LLMs actually replace them. What they find is that LLMs can speed up the parts of the work that are tedious, and that they can create a host of issues that later need to be fixed by a competent human.
I suspect that whether an LLM actually can replace an individual engineer or programmer may depend on whether that person is in the 20% that gets 80% of the real work done, or in the 80% that can only get 20% done.
But on the gripping hand, is do the CFOs think that they can replace their engineers or programmers with LLMs and still collect about the same revenue? And that’s the real issue, not can the machines replace the craftsmanship, but will enough of the C-suite believe that they can well enough to still make money?
Niketas,
I can think of no better, real world application for LLMs than computer coding. Of course they will replace human programmers. The more rules and constraints a system has the easier it is for an LLM to emulate “human thinking.” This is why computers cracked the code on Chess sooner than they did on Go. Chess has a lot of rules and is very rigid. Go has few rules and is very flexible.
A spoken language, like English, has a lot of grammar rules and is constrained by a relatively small, finite number of words. So mimic’ing human speech isn’t too difficult, but mimc’ing human thought is.
When writing a computer program a software developer has a defined goal and the language he or she chooses to use has a finite, tiny vocabulary and a stringent set of rules that must be followed. There is no; “to, too, two” in a computer language, or “there, their, they’re.”
If a friend walks up to me and says, “Make me laugh.” I can likely do it. I know my friend and I know what makes him laugh, but an LLM would have to spend time learning about my friend, what makes him laugh, how does he differ from other humans, what episodes in his past can be drawn on for humor, what can I gather from his physical demeanor, posture, facial expression right now that may be different from when he laughed yesterday?
But prompt an LLM to use Python to create an app that has an input field where a user can log purchases and their amounts, along with a date time stamp, and track them in a database against a monthly budget? Easy-peezee. Compared to all the potential ways to make my friend laugh, including nonsense sounds that are not in the English language, the world of Python is very easy to maneuver.
@Rufus:But prompt an LLM to use Python to create an app that has an input field where a user can log purchases and their amounts, along with a date time stamp, and track them in a database against a monthly budget? Easy-peezee.
I’d be very surprised if LLMs were saving engineers’ development time on easy stuff, but I’m not a professional developer. What’s at this link is long but might give you a better idea what I’m talking about.
1. Prompt Debt
This is the most visible form of AI debt, and often the first that teams encounter. It starts innocently. You write a prompt that works. Then you discover an edge case, so you add a clarifying sentence. Then another edge case. Then someone else adds a constraint you didn’t know about. Six months later, your prompt is 800 tokens of nested instructions that nobody fully understands….
2. Model Dependency Debt
Many organizations are building on models they don’t control. You call an API, you get a response, and somewhere in between is a model that might change tomorrow.
This creates several debt patterns. There’s vendor lock-in — you’ve built features that depend on provider-specific capabilities that don’t transfer cleanly to alternatives. There’s API stability debt — you’re relying on endpoints or behaviors that might be deprecated. There’s model version debt — your system works on GPT-4-turbo-preview but breaks on the latest version because output formatting changed. And there’s pricing model debt — you’ve built something under current pricing that becomes economically infeasible when rates change….
3. Data Pipeline Debt
LLM systems rarely work in isolation. Production deployments often use Retrieval-Augmented Generation (RAG), retrieving relevant information from knowledge bases and providing it as context. This creates a category of debt around data pipelines.
RAG quality debt accumulates when retrieval systems degrade. Maybe your chunking strategy — how you split documents — was optimized for your initial use case but doesn’t generalize. Maybe your embeddings are stale — you generated vector representations six months ago and haven’t refreshed them as models improved. Maybe your knowledge base contains outdated information that nobody’s systematically reviewing….
4. Evaluation Debt
This might be the most dangerous category because it makes all other debt invisible. If you can’t measure output quality, you can’t detect regression. And LLM outputs resist simple measurement….
5. Observability Debt
Production LLM systems are remarkably opaque. A user submits a query, something happens inside a black box, and a response emerges. When something goes wrong, you have limited visibility into why.
Observability debt accumulates as missing monitoring. Can you detect hallucinations — outputs that are plausible but factually wrong? Can you track model drift over time? Can you identify which queries are costing you money? Do you have audit trails for compliance requirements? Most organizations answer “no” to several of these questions….
6. Integration Debt
LLM systems don’t exist in isolation; they integrate with existing enterprise infrastructure. And those integrations accumulate their own debt.
There’s glue code debt — excessive middleware and adapters connecting LLMs to other systems. There’s orchestration debt — complex workflows coordinating multiple models or tools that become brittle. There’s error handling debt — inadequate failure management that leads to cryptic user experiences when things break. And there’s state management debt — conversation history and context persistence that works fine for early users but doesn’t scale….
7. Governance Debt
As LLM adoption accelerates, organizations discover they lack basic governance structures. Nobody’s quite sure who can deploy what models, what the approval process is, or what policies govern model usage.
Governance debt manifests as compliance gaps — regulations require certain controls that aren’t in place. It shows up as access control debt — permissions are ad-hoc rather than systematic. It appears as missing model approval processes — teams are shipping LLM features without vetting. And it accumulates as policy documentation debt — usage guidelines are tribal knowledge rather than written policy….
What I am not particularly fond of is all of the videos where “animal lovers” are kissing the animals they have found/rescued–possums, racoons, dogs, rats, mice, chinchillas, pigs, ducks, parrots, wombats, iguanas, various monkey species, alligators, donkeys, etc., etc.–which they declare to be the “love of their life” or some such, and proceed to keep kissing them on their mouths.
Well they don’t rescue Dung Beetles, or do they?
Nonapod: “a then unglaciated Antarctica”
Due to Flintstones’ SUVs …
Nik: “I find possums revolting”
I bet none are so committed to the cause that they have attacked any ICE personnel.
Cute video—I never knew that playing dead kicked in automatically for those marvelous marsupials.
Nonetheless, should opossums be upstaging groundhogs on February 2nd, I ask you!
A few months ago I chatted up a nice older gent on our flight to Dallas. So Texas! He showed me all kinds of pictures of his small hobby ranch and their chickens and goats and dogs and cats. Showing me a photo taken inside their house, he was trying to figure out which of their dogs was curled up in a bed in front of the fireplace. I took a look and laughed and said that I couldn’t get a good look but haha it looks like an opossum.
Of course, it was. The wife had taken in two of them, actually. They like to come in and sleep, get fed, and then waddle back outside for the day. Cute videos, too.
Our English Springer Spaniel has caught a few opossums. She has the hunting dog’s “soft mouth,” so she grabs them but doesn’t break the skin. She proudly brings them to us and drops them outside in exchange for a treat. We bring her in, and in a few minutes the possum will cautiously look around and then waddle away. They really do play possum!
Does the possum get a treat, too?
Speaking of treats…
Time to play Compare and Contrast:
The Hannecke article, which Niketas linked, describes using AI to generate software purely by prompts, which sounded so insane to me I couldn’t believe professionals were using AI that way.
Mostly they aren’t. Instead they use AI as a junior developer. They give AI a well-specified prompt for some code. They review the code, integrate it into the codebase, run it, test it, fix as necessary. Lather, rinse, repeat.
AI will get better and it will be able to handle increasingly more complex programming tasks on its own, but we are a long ways away from “Prompt it, forget it.”
I was way behind in my reading but I caught up yesterday.
@ huxley (on 1/28 “Yesterday”) – “I was glad I didn’t learn what cataract surgery was until after the fact. A full description sounds fairly insane on paper.”
So true – I had both eyes done about 15 years ago (!!) and am glad I did, but as insane as cutting open an eye is, I think cutting open a skull is a lot crazier.
Although I believe I read once that the ancient Egyptian doctors would do it.
Anyway, I hope you are feeling better and have adjusted to your new visionary apparatus.
I will check in later this week, and let you know about my trip to the hospital. Sadly, Dr. Kildare will not be in attendance.
JEFFREY EPSTEIN LAST INTERVIEW FROM HIS HOME (from Epstein files) Steve Bannon [Jan 31 2026] Full Video
I generally give possums a pass, even though I have to liberally dose my bird seed with cayenne pepper to keep them out of the feeders. But their low body temperature means they can’t contract rabies, and even better, they eat ticks. Lots and lots of ticks. So I give them a pass, for the benefit of the rest of us mammals.
And another so-called “conspiracy theory” proven to be fact(!)…thanks to the morally-bankrupt, utterly-corrupt, Socialist/Communist (but I repeat myself, of course) governing coalition in Spain….
To be fair, population replacement (and the tremendous political power and opportunities for grift such a “policy” is intended to provide) is also enthusiastically supported by the DPUSA, the EU and the Anglosphere (with the exception of America’s current leader and certain still-sane Eastern European and Latin American countries)…
Off target, but after seeing Billie Eilish (butt ugly intentionally) at the Grammys I think there is a lot of money to be made with urinal targets.
Intentional black and white attire, the hate filled countenance, sells itself. Really tops the 1960s Kane Fonda …..
Leave a Reply
HTML tags allowed in your
comment: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>
I never knew a possum was a marsupial.
Live and learn (but only if you choose too).
Judging by YouTube there are quite a number of people who take in young animals who’ve lost their mother and would die without help.
They feed and raise these little creatures until they are mature enough to be released back into the wild. Sometimes they leave, sometimes they keep coming back.
Here’s a young woman who took a baby opossum in. They bonded so thoroughly that the opossum would always come back. She has accepted that they will be together.
–The Dodo, “Opossum Covers His Mom’s Face In Kisses”
https://www.youtube.com/watch?v=6VW73ZFZiLI
Cute li’l guy!
It’s believed that marsupials originally evolved in South America somewhere between 125–160 million years ago and migrated to Australia through (a then unglaciated) Antarctica via the Antarctic land bridge that existed for awhile during the Cretaceous period. They really couldn’t compete with placentals too well, so they only persisted in South America and Australia, both of which remained isloated from the other continents until South America came into contact with North America sometime around 15 million years ago.
Yes I know the physical difference between a o’ possum and a possum, still they’re all possums to me.
I didn’t even know that there were possums in Ireland!
And they are immune to rattlesnake venom.
Madison Kanna wrote at X: ‘as a software engineer, i feel a real loss of identity right now.
for a long time i defined myself in part by the act of writing code. the pride in a hard-earned solution was part of who i was. now i watch AI accomplish in seconds what took me hours. i find myself caught between relief and mourning, awe and anxiety. the craft that shaped me is suddenly eclipsed by a machine. who am i now?’
Her post inspired a post & discussion about skill obsolescence in historical perspective:
https://chicagoboyz.net/archives/76039.html
I personally find opossums revolting…
Madison Kanna wrote at X: ‘as a software engineer, i feel a real loss of identity right now….i watch AI accomplish in seconds what took me hours….
There are plenty of software engineers who don’t think LLMs actually replace them. What they find is that LLMs can speed up the parts of the work that are tedious, and that they can create a host of issues that later need to be fixed by a competent human.
I suspect that whether an LLM actually can replace an individual engineer or programmer may depend on whether that person is in the 20% that gets 80% of the real work done, or in the 80% that can only get 20% done.
But on the gripping hand, is do the CFOs think that they can replace their engineers or programmers with LLMs and still collect about the same revenue? And that’s the real issue, not can the machines replace the craftsmanship, but will enough of the C-suite believe that they can well enough to still make money?
Niketas,
I can think of no better, real world application for LLMs than computer coding. Of course they will replace human programmers. The more rules and constraints a system has the easier it is for an LLM to emulate “human thinking.” This is why computers cracked the code on Chess sooner than they did on Go. Chess has a lot of rules and is very rigid. Go has few rules and is very flexible.
A spoken language, like English, has a lot of grammar rules and is constrained by a relatively small, finite number of words. So mimic’ing human speech isn’t too difficult, but mimc’ing human thought is.
When writing a computer program a software developer has a defined goal and the language he or she chooses to use has a finite, tiny vocabulary and a stringent set of rules that must be followed. There is no; “to, too, two” in a computer language, or “there, their, they’re.”
If a friend walks up to me and says, “Make me laugh.” I can likely do it. I know my friend and I know what makes him laugh, but an LLM would have to spend time learning about my friend, what makes him laugh, how does he differ from other humans, what episodes in his past can be drawn on for humor, what can I gather from his physical demeanor, posture, facial expression right now that may be different from when he laughed yesterday?
But prompt an LLM to use Python to create an app that has an input field where a user can log purchases and their amounts, along with a date time stamp, and track them in a database against a monthly budget? Easy-peezee. Compared to all the potential ways to make my friend laugh, including nonsense sounds that are not in the English language, the world of Python is very easy to maneuver.
@Rufus:But prompt an LLM to use Python to create an app that has an input field where a user can log purchases and their amounts, along with a date time stamp, and track them in a database against a monthly budget? Easy-peezee.
I’d be very surprised if LLMs were saving engineers’ development time on easy stuff, but I’m not a professional developer. What’s at this link is long but might give you a better idea what I’m talking about.
We are screwed indeed.*
* See https://www.youtube.com/watch?v=t7Cfb_hLoRI
Some might find this description of rewriting NumPy in Rust using Claude interesting.
RE: Possums and other critters–
What I am not particularly fond of is all of the videos where “animal lovers” are kissing the animals they have found/rescued–possums, racoons, dogs, rats, mice, chinchillas, pigs, ducks, parrots, wombats, iguanas, various monkey species, alligators, donkeys, etc., etc.–which they declare to be the “love of their life” or some such, and proceed to keep kissing them on their mouths.
Well they don’t rescue Dung Beetles, or do they?
Nonapod: “a then unglaciated Antarctica”
Due to Flintstones’ SUVs …
Nik: “I find possums revolting”
I bet none are so committed to the cause that they have attacked any ICE personnel.
Cute video—I never knew that playing dead kicked in automatically for those marvelous marsupials.
Nonetheless, should opossums be upstaging groundhogs on February 2nd, I ask you!
Well, possibly…O-Walzums…
https://tinyurl.com/zj2knbwf
A few months ago I chatted up a nice older gent on our flight to Dallas. So Texas! He showed me all kinds of pictures of his small hobby ranch and their chickens and goats and dogs and cats. Showing me a photo taken inside their house, he was trying to figure out which of their dogs was curled up in a bed in front of the fireplace. I took a look and laughed and said that I couldn’t get a good look but haha it looks like an opossum.
Of course, it was. The wife had taken in two of them, actually. They like to come in and sleep, get fed, and then waddle back outside for the day. Cute videos, too.
Our English Springer Spaniel has caught a few opossums. She has the hunting dog’s “soft mouth,” so she grabs them but doesn’t break the skin. She proudly brings them to us and drops them outside in exchange for a treat. We bring her in, and in a few minutes the possum will cautiously look around and then waddle away. They really do play possum!
Does the possum get a treat, too?
Speaking of treats…
Time to play Compare and Contrast:
Brendan O’Neill—
“This ayatollah fanclub heaps shame on London;
“As Iranians fight for freedom, people in London praise their theocratic murderers.”
https://www.spiked-online.com/2026/02/02/this-ayatollah-fanclub-heaps-shame-on-london/
(I’m pretty sure that neither Starmer’s government nor London’s is capable of feeling shame.)
“New York Times Misleads Readers on Gaza Death Toll”—
https://freebeacon.com/israel/new-york-times-misleads-readers-on-gaza-death-toll/
H/T Powerline blog (for both).
(NYT actually just doing what the NYT does…)
And this bombshell:
Who knew?
Woody Allen’s “Sleeper” is actually a serious documentary on Bernie Sanders disguised as a literary comedy.
Oh…wait…
“How Bernie Sanders built a device to give himself ‘cosmos-shattering orgasms.’”—
https://instapundit.com/773701/
File under: The Reich Stuff
+ Bonus (blood hounds, AKA Inner Space)…
Some might find this of interest:
“Scientists Identified a New Blood Group After a 50-Year Mystery.”—
https://instapundit.com/773470/
Re: AI / programming
The Hannecke article, which Niketas linked, describes using AI to generate software purely by prompts, which sounded so insane to me I couldn’t believe professionals were using AI that way.
Mostly they aren’t. Instead they use AI as a junior developer. They give AI a well-specified prompt for some code. They review the code, integrate it into the codebase, run it, test it, fix as necessary. Lather, rinse, repeat.
AI will get better and it will be able to handle increasingly more complex programming tasks on its own, but we are a long ways away from “Prompt it, forget it.”
I was way behind in my reading but I caught up yesterday.
@ huxley (on 1/28 “Yesterday”) – “I was glad I didn’t learn what cataract surgery was until after the fact. A full description sounds fairly insane on paper.”
So true – I had both eyes done about 15 years ago (!!) and am glad I did, but as insane as cutting open an eye is, I think cutting open a skull is a lot crazier.
Although I believe I read once that the ancient Egyptian doctors would do it.
Anyway, I hope you are feeling better and have adjusted to your new visionary apparatus.
I will check in later this week, and let you know about my trip to the hospital. Sadly, Dr. Kildare will not be in attendance.
JEFFREY EPSTEIN LAST INTERVIEW FROM HIS HOME (from Epstein files) Steve Bannon [Jan 31 2026] Full Video
https://commoncts.blogspot.com/2026/02/jeffrey-epstein-last-interview-from-his.html
@AesopFan:Although I believe I read once that the ancient Egyptian doctors would do it.
Done from Stone Age times all through the Middle Ages to the 20th century. Nothing experimental about it, it’s tried and true. Anyway, good luck.
I generally give possums a pass, even though I have to liberally dose my bird seed with cayenne pepper to keep them out of the feeders. But their low body temperature means they can’t contract rabies, and even better, they eat ticks. Lots and lots of ticks. So I give them a pass, for the benefit of the rest of us mammals.
And another so-called “conspiracy theory” proven to be fact(!)…thanks to the morally-bankrupt, utterly-corrupt, Socialist/Communist (but I repeat myself, of course) governing coalition in Spain….
(What a crew!)
“…Far-Left Spanish MP’s Demand 500,000 Illegals Be Made Citizens For ‘Population Replacement’”—
https://www.zerohedge.com/geopolitical/advocating-genocide-musk-slams-far-left-spanish-mps-demand-500000-illegals-be-made
To be fair, population replacement (and the tremendous political power and opportunities for grift such a “policy” is intended to provide) is also enthusiastically supported by the DPUSA, the EU and the Anglosphere (with the exception of America’s current leader and certain still-sane Eastern European and Latin American countries)…
Off target, but after seeing Billie Eilish (butt ugly intentionally) at the Grammys I think there is a lot of money to be made with urinal targets.
Intentional black and white attire, the hate filled countenance, sells itself. Really tops the 1960s Kane Fonda …..