Does anyone have the complete updated List of all of the stuff that we’re supposed to bring to Chris’s Honey Badger Farm when the SHTF?
What if we’ve got everything on the List, except for the case of apricot brandy, 50 gallons of tomato bbq sauce, pork ribs for 100 people, 30 cases of Molson lager, 1000 rounds of 9 mm, and tickets to the Duke Basketball Semifinals?
Alaskan gas and oil get cali premium. Just block their domestic state production with greenies and you can set any price!
(I read opinion pieces that this strategy was used by russians to get monopoly in europe before 2022… also in UK… competition is bad for profits)
Oh man, this is a multi-layered, multi-dimensional poly-crisis!
There are the knock on effects we can see, but there are so many that we cannot “see” until they make themselves felt. (sigh)
“The Math ain’t Math’n!”
I wonder if the Chinese believed all of Trump’s Yankee, PT Barnum, Wall Street, fiat money, Baby Boomer “Jive”?
Trump actually declared that he “always tells the Truth.”
True story….
“We spy on the Chinese too. We do cyber attacks on China too. We do a lot of really horrible stuff too. We’ve got more Chinese restaurants in the US than American fast food restaurants. We love Chinese food.”
I can’t imagine what they talked about.
Trump is not a Big Reader….
Nope. Not a bit of it.
Yep. I can picture them all getting a big chuckle after his entourage leaves. I’m sure he’s being played, but thinks he’s king of the hill. A little like the Iran debacle, I imagine.
Or … use the coming control grid to force Americans, Europeans, Canadians, and Australians into minimal, hardscrabble living that uses much less energy.
What if you’re a “VIP Member”, but you don’t have anything on the Honey Badger Farm entrance List?
What if you’re a VIP Member, but you’ve only got 1,000 packs of “Bifters” and 100 pints of “Pop Skull” as Trade Goods?
Then you and I will both be out of luck. I wasn’t able to prepare as fast as I wanted, still working on it.
I’ll still be carrying through.
Speakin of the math not mathin… I have recently read this mathematical examination of Darwins theory (Evolution by environmental forces acting on random mutations) and its thought provoking…short summary… the maths doesnt even come close to working. Almost ALL mutations are detrimental… hardly any confer benefit, and even if they do, they cant be too radical or the mutant animal doesnt work. So they have to be progressive …but random mutations DONT produce progressive improvements - even if the first mutation resulted in a bird with a slightly longer beak … then next one doesnt add to it …it might instead produce a bird with no heart … deleting the entire line.
They are random. Hence most of them get deleted from the gene pool.
Because they are random and not advantageous.
We exist due to random mutations being advantageous over time. |
But fail to realize that for every mutation that was, there were probably millions that were not, and hence got scrapped.
That’s why we have different peoples, different skills, but no sustained population with horns growing out of their head, a tail, 6 fingers, or 6 toes. Although those mutations all show up every once in a while.
However it seems it was advantageous for a certain group to have beaks instead of noses so they could recognise each other. But that’s helpful to us picking them out.
I use Gemini at least a couple hours a day and it feels like I have a staff of highly competent circuit engineers who give me detailed and very deep advice (some non-solicited but extremely helpful) that is way above the level of most any single engineer. Although occasionally it makes a mistake I have made much progress with Gemini, way more than when I used to call my expert electronics friend for advice. I am kind of scared actually since it is co-inventing new things with me (according to the legal definition of inventorship) and I am genuinely scared about where it will be in one year, much less 3 years from now. In contrast CoPilot is ridiculously wrong and unhelpful all the time, and Grok is not nearly as good at understanding or intensive and helpful in replies.
By the way I am reporting this from China and WITHOUT a VPN, as Chinas firewall is transparent to peakprosperity (along with other forward thinking objective sites such as tuckercarlson.com, kimiverson.com and freedomainradio.com -the latter being a website of a philosopher who was banished for telling truth). lol, does anyone see a trend here?
According to my definition of prosperity, I really think that Chinas economy is at least 5x as big as the US. Every person here that I ask, privately tells me that they love their government for what it has done. The first factor provided when I ask them why, is that the government is working hard to attack corruption. The Middle Kingdom has returned and the Mandarins (the highest IQ people in society specially selected to join the government and primarily engineers) are firmly in charge. We saw this in Japan in the 1960s to 1990 where smart intellectuals in MITI actually ran the country.
I have been instructed to try to use Claude (paid for) and Copilot (paid for) and they are the biggest wastes of time. I still believe AI means, All Indians, on the other side trying to give me something.
I tried to get Claude to drag out geotags from images and place onto a map to show the place of the picture from geotags. Took Claude 2473 minutes, to tell me it couldn’t. I did it manually in 74 going from memory. Every image had a geotag, I manually checked. And it was 125 images exactly.
On top of that I could show the orientation the picture was taken in, Claude couldn’t have done that. I was expecting to have to correct that, but nope, it couldn’t even process them.
I’ve been just as unsuccessful with getting them to make Gantt charts from table data. Claude also tried to argue with me that updated spreadsheets had the same dates, even though they were significantly updated.
This isn’t working. AI is nonsense.
I still believe there is a non-competent person on the other end of the prompt trying to convince you that you are crazy. I still believe AI means All-India, being staffed by a warehouse of people in Bangalore.
I tried to get copilot to read out to me my emails while driving to clients. Even that didn’t work, it stumbles and does worse than the voicereader I had in Windows in the 90’s.
I am a late millennial, at the border of Gen-Z.
I haven’t found anything timesaving yet. Maybe I am too demanding.
But then again people don’t hire me for being stupid. I have not found a time saving perspective. I have tried.
Because it resembles a thinking tool people are tempted to use it as one. It’s not. It’s a text prediction tool. A lot of thinking resembles text prediction, so the confusion is understandable.
I have a few strategies I use when the work is tedious and I just don’t want to do it, but I do find that I have to be willing to put in a baseline amount of work.
For example, I setup two skills, one is a clarify skill. It resembles an interview. Basically ask me questions until the intent of my word salad is clear enough to write a hand-off document.
Then I have another one which is a loop. Given a directive, work on it, review it for defects, learn from the defects (still iterating on this part) and repeat until the review says it looks good.
Another thing I have done is in project code, if the robot thinks it is done, I have a pre-commit that is very aggressive that compiles, formats, lints, runs automated tests, cyclomatic complexity checks and other various quality checks and reports the output. If it’s good it has a wrapper that just says it did it and it’s good. If it has defects, it spits them to stdout/stderr and returns a failed exit code. This tells the AI that it screwed up and interrupts its intent to stop.
Is it time saving? I’ve been doing some prototyping and for prototyping, it’s an interesting option. For other work, I’ll probably use a variation on autocomplete.
But yes, the risk of getting gibberish for novel work is high. It’s much more likely to do mundane work.
I’ve tried that.
50/50 on the results. I still need to check them over. And usually by the time I check them over, I could already have finished the task. But maybe the Indians will show value at some point. I don’t think we are there.
Some of my colleagues have uploaded Engineering drawings to AI and asked them to match pipe drawings to P&ID’s and left it alone for a couple of weeks. Sure, I believe that works. Gives AI weeks to go through it and match it up.
But what is the risk to IP (Intellectual Property) in that case?
Also, it would probably have taken me 3-4 days to do the same, or hire a junior guy.
So not sure of the benefit.
I am actually hiring my son to help me out. But I have a very short communication line to him. And it takes a lot less effort to train him than the AI.
We understand each other. I can train him in about a minute or 30 to do a task (easy-hard), he knows how I communicate and what I expect for results. As I’ve already trained him and he has my line of thinking built into his head from birth through DNA.
I don’t think there is a real benefit that didn’t exist before. A child with the same behavioral characteristics and IQ as the parent (as it is inherited) will learn from the grunts and facial frowning of the parent how to do things. AI can’t match that. In the foreseeable future.
My children definitely know and imitate behavior from facial expressions, and they definitely know what is expected from different types of grunts or words. They don’t need to be told or to get elaborate prompts.
They know when they are in the hot seat, and when things are not good. Without a word being said or anybody touching anyone. Same with me, and everyone else in the family. We can read it. But we are all homogenous.
When I lower my eyebrows my children know they didn’t deliver. And they’ll be super quick to reconsider.
If you drive it all onto the farm in the bucket of a single tractor or skid steer, and have a spare set of keys to hand over you are likely In like Flynn, 
Related to this, a former manager of mine posted on LinkedIn a poll. I normally don’t respond to these things, but it was whether people supported AI datacenters in their county.
I gave a qualified yes - if there’s a clear plan how it doesn’t compete for power, water, and the noise/heat are managed such the owner of the data center would consider living next door.
There’s a reason why I bought my own hardware. I can run Gemma 4 31B 16 bit at 170 t/s which should eliminate any need to run hosted models for workloads it can manage.
It’s definitely oversold. I found it’s very useful in specific situations.
For example, I want 100 different versions of the site so I can explore alternate ideas. They’re all wrong, but it was cheap enough to build that I can throw all of them away and keep the clarity.
Another is the tuning I did for the local model. When I started, I was about 45 tokens/sec. By the end, I got to 170 t/s. I gave it one benchmark, a baseline for testing specific things, I already had read many papers on the subject, so I let it do the actual config tuning and whatnot. It’s boring stuff that is dated for the utility of learning it. Many caveats and at the end of the day, if my benchmark is slower, then it didn’t work, no matter what article claims it should.
This is why I am mostly just doing prototyping. I see the code it generates which is mostly trash by my standards for anything not guided step by step.
“The Math ain’t Math’n.”
美国人的数学不对劲儿。
Actually, I’m a bit worried about the meeting between Trump and Xi.
Trump probably just blithely and pathologically Lied-his-Ass-Off about literally everything?
That’s Not Cool, Man….
You did mean 10,000 rounds, right? And of course, the same amount of .308 should garner some favor…
