ThunderSizzle

Also, if you then resume your conversation, and the LLM discovers the change, it might revert the "accidental" change.

If your working with an active context, and changes you do then need to be conversed back to the agent, and even then, it might still find it jarring and wrong.

The ability by the supplier is the problem, regardless of there is an upfront clause or not. The upfront clause is the problem

If Ford told the DoD you can't use the tanks they bought however they need to, and had a kill switch in it in case they wanted to cut the DoD off, then that would obviously be a supply chain risk.

The military and it's suppliers need mitigate risk, and that includes suppliers intentionally causing harm to soldiers on the front lines, or in the backend logistics supply chain.

Honestly, every company should be treating Anthropic and OpenAI as supply chain risks, and should completely avoid dealing with them.

I think the answer is the military finds another vendor willing to deal, or move it in house.

It's not like Anthropic will be nationalized any more than the bankrupted auto companies of 2008.

a flight attendant was annoyed that my diaper bag was in front of feet instead of in front of my feet by 3 more inches. While feeding a fussy baby as a parent traveling alone, with no hands free, I gave then a look of "are you serious?"...while I proceedes to make my daughter cry quite a bit to shove the bag a bit more.

Just being a flight attendant doesn't make them great humans to be around.

I did list GDP growth per capita[0], so that already was accounting for population changes.

Having said that, you are depressingly very correct. A negative population growth will cause the post-2008 depression to continue even longer, and probably more agressively.

[0] https://data.worldbank.org/indicator/NY.GDP.PCAP.KD.ZG?end=2...

Did you try other models (e.g. Qwen3.6 or others)? I'm curious how others fare.

I've noticed Qwen3.6 struggles a bit with today/date based logic.

The country has been _run_ by fools for 26 years. Congress has had 26 years to do something about the fiscal situation, and we've had four presidents, and the fiscal responsible side of the electorate is never listened to.

Both sides are to blame - neither will fix the problem. Obama could've made that his goal - he was competent, had a lot of political good will, and many people were frustrated at the bailout policy Bush did, but instead it was inflationary printing (quantitative easing), Obamacare and Cash 4 Clunkers (which the used car market still hasn't recovered from).

I never voted for him - I didn't view him as honest, nor did he seem to indicate that he liked America, but was rather just a good talker - but I think he could've been a great president given a less radicalizing agenda.

He was probably the best situated president in terms of timing to fix the debt problem, but instead it was a good time for divisive politics. By the time Obama finished, it became clear neither party actually cared about the fiscally conservative Ron Paul supporting voting block.

We'll continue through the depression we've started since 2008. (GDP growth should be closer to 3.5%-5%, but we haven't really escaped sub-2% since 2008) - our GDP has been depressed by at least 1-2% growth since that crisis, and I think a large part of it has been the inflationary cycle we started and never stopped.

The wars already put us into too much debt, Obama continued it for 8 years (granted, the deficit slowly went down, but it wasn't fixed). Trump and Biden did a huge disservice to the debt (but neither really cared much about it), and now I fear the path Bush, Obama, Trump, and Biden have laid will not be easily fixed.

Well, I don't see a value issue of using Qwen3.6 27B vs Sonnet 4.6 (not sure about 5 yet)

I still have to use GHCP at work, and I self-host at home, and aside from the fact self-hosting also forces you to tinker, optimize, etc. - there's not a huge difference in my end result in end user results. I spent quite a bit of time trying to optimize llamacpp and compare 35b to 27b, etc. I don't compare models that much at work.

I guess the other part of it is I didn't really know much about cheaper cloud models, but I was attracted to the idea of no longer renting against Claude code, etc. I figured if I could run something functionaly similar from my bedroom on a normal outlet, then all this talk about data centers needing to be built everywhere in the news cycle is obviously just plain stupidity and hype.

It appears I'm using about 20.4/7.6 million in/out tokens a month, or on open router, about $20/month.

That puts $1350 at a 5-6 year break even (thanks to cheap electricity), I guess. Beyond that, running on localhost as a nice feature of 0 no latency when doing rapid tool calling

Sadly. It's jumped $300-400 in 2-3 months time.

I don't think I'm the only persona that did the math.

Claude Code is $100+ or else be constantly throttled. My usage on GHCP was gonna be $300+ a month.

I paid $1350 and threw an R9700 in an existing machine. That's a 4 month pay off or so.

Plus, I can feed it sensitive data all day and not be worried where it's going.

Weather forecasting is important even before solar/wind. Extreme, temporary temperature swings require the power company to be prepared for the extra power generation.

With just coal/gas/hydro, this can be mitigated much easily, but still needs to be handled with care. If not, then the lack of power generation could cause brownouts for hours or days as backup power generation takes time to spin up.

TVA experienced an issue like this in 2023, where it was inadequately ready for a winter storm: https://www.publicpower.org/periodical/article/report-examin...

Since we build for "1-in-a-100", it's expected we'll have these failures, as a 1% chance every year is a pretty high chance that will keep increasing as these power companies rely more on non-failsafe power generation strategies. Texas is the extreme example, where over-reliance on those systems caused too much damage.

Interesting perspective. I toyed with rust a bit from the perspective of a c# background (and a bit of java, php, classic asp, JavaScript, typescript, etc) .

I like rust, but I've come to still prefer c#'s object oriented features. Perhaps it's my naivety, but I've found c# AOT compilation to do plenty of trimming and startup performance optimization that I don't see it as a bad option.

Have you made a personal comparison on Rust vs Object Oriented Languages like c#?

I have backup cards. If one gets stolen, I can switch to a backup card instantly.

If my phone is lost, then apparently I need to buy another one with the...phone I don't have? A single point of failure is horrible design.

Strawman, but you raise a good point. Credit card terminals can still work in certain disaster situations more reliably than trying to get everyone to not lose their phones and keeping them charged and wasting precious energy on a transaction.

A credit card doesn't require electricity, and even the classical process of carbon copying can work without power or internet to the pos (though that has its own problems - but we're talking "disasters")

Regardless, a person not having a smart phone should not be grounds of exclusion from society. It's too vulnerable of a device for it to be a single point of failure. It already is too critical of a device even for those that have one, that losing it is already a very big burden. Even just switching to a new phone via the migration process on Android can still be a burden.

No, but it shows up in the bill just like everything else.

Cost of doing business

What happens if you don't have a phone that supports the App, or if you forgot your phone or lost it (aka stolen)?

How are you allowed to function if the App is missing?

If you use it, that's great. If you don't use the gift card, then cash is obviously better

Well, in my state, you don't "apply" for a gun like you apply for a license. It is just document that you purchased it after the seller runs a background check.

The government isn't attesting to your capabilities - that is still on you as a citizen.

Corruption has been rampant under multiple administrations. I actually had hoped Trump might root out more corruption in his 2nd term, but two things happened: Congress wouldn't go afraid the corruption that was being found, and Charlie Kirk was assassinated in broad daylight...after 2-3 assassination attempts on Trump himself too.

It's easier to start looking the other way when bullets start flying.

Having said that, I'm not saying Trump was a saint, but it seems this board forgets we just had Obama and Biden running their corruption rings for 2 decades too...

I still run with u matrix though, and most third party requests can be limited, but I no longer have enough patience when a required page doesn't work - I'll just rely on ublock origin to work.

Right now, at least, I haven't found an AI capable of replacing a software engineer. Ive seen AI that can easily replace basic programmers, however.

I also think AI can replace non-coding artitects, and probably most middle management type jobs (my company has 6 levels of management between the CEO and "individual contributors" in my area (8 counting inclusivity)).

That's a lot of management levels, and every level has to be paid more than the level they manage as a fraud disincentive. So that's a lot of money...

I use comments for external domain constraints (e.g. this table has three types of records in them that we use for different purposes - I might have the LLM agent be a bit thorough and clearer with the explanation, but I found those type of comments very much helped future iterations)

I think MAGA would be okay with the...fly over provinces..is that what they would be called?

The rest is too woke, too British, and too CCP oriented to work with.

Most code I write these days are never directly consumed by anyone or anything. Daily or monthly batch jobs that help keep the data flowing correctly.

Ive made SPAs in Blazor that make it a point to update the URL whenever a "navigation" occurs. I also made it a point to ensure refreshing or loading the URL restored the state.

It's a small problem compared to the ease of use for development you can gain via Blazor, so I don't see any reason not to solve it.

Well, they also need to ensure they aren't spending past their premiums coming in...

Alot of insurers the past few years have actually gone bankrupt from ballooning costs and claims.

So, they want costs to go up, so the next year they can up the premiums, so they can then up their admin spend, but the costs still need to below the premium. And they can't just up the premium without an associated rise in cost, because groups are owed their premium money back if it isn't paid out.

It's even worse when they claim to be "in-network" and they make you sign that. To me, the entire point of being in network is I shouldn't have to worry about the provider filing claims correctly on my behalf, and if there is an issue, it should be on the "network" side, not my side.

Insurers can only spend 15% of premiums on admin spend (e.g. salaries, etc). That means if the CEO (and everyone else) wants more money, they need to pay more money to the hospitals, so their 15% cut grows.

Also, for beyond that, unspend premiums have to be refunded, which also cuts into the potential admin spend bucket, so it's best to always ensure your paying as much as possible in claims.

Plus, patients hate denied claims, because typically your forced to sign paperwork saying your responsible for anything insurance refuses to pay.

And with most health insurance spending being locked to employer benefits, the big market is the group market, which means your stuck with either getting your job's option(s), finding a new job, or not getting insurance basically.

In other words, you don't have a price choice as an individual in a group.