Wall Observer BTC/USD - Bitcoin price movement tracking & discussion

[WatChe]

It still seems retarded to me, and I am standing by my earlier assertion, whether you end up being correct or not remains almost irrelevant...

And, of course, even if there were a bet in there, you may well continue to NOT be a betting man.

hahahahaha
  

Well for betting, I will just reiterate my previous stance which is "in gambling, the winner is the one who knows when to stop".

Its better to stay away from gambling because one of participant has to lose money.

Between What are results of your latest bet on WO?

[ChartBuddy]

Explanation
Chartbuddy thanks talkimg.com

[Hueristic]

Dude.  You write so sparingly yet take the time to enter the alt code for an archaic em dash—why?

Ah, so I am not the only one who has noticed.

While other similar experiments seem to have run their course or at least have been put on pause,

If being lazy and using standard phone keyboard shortcuts makes me an "experiment," you guys are overthinking this way too much, the tin foil hats are definitely coming out today.

For shame, are you making me feel like an idiot for giving you so much credit so early?

No-one is going to believe you typed that and did not copy and paste it.

Unless you explain exactly how you managed it.

[JayJuanGee]

knock knock who is there

Eighty thousand for right now

but what is  the top?

There is no top.

OOM beat me to it.   

It still seems retarded to me, and I am standing by my earlier assertion, whether you end up being correct or not remains almost irrelevant...
And, of course, even if there were a bet in there, you may well continue to NOT be a betting man.

hahahahaha
  

Well for betting, I will just reiterate my previous stance which is "in gambling, the winner is the one who knows when to stop".
Its better to stay away from gambling because one of participant has to lose money.

Between What are results of your latest bet on WO?

I might have proposed some bets, yet I cannot recall entering into any bets since my lost bet with LFC.. so it has been a while with no bets.

[OutOfMemory]

Soon: ETF daily settlement time windows. I'm curious if there will be another push up today.
Had my last buy yesterday around $67.50076.500.

[goldkingcoiner]

Did the news of Saylor sitting next to Jeff Bezos really send Bitcoin flying? Lol.

Also a lot of you guys are getting way too comfortable with giving AI total control.

[cAPSLOCK]

I'm kind of a no-trusty type person.

In fact, I would encourage you not to trust me a single bit.

It's a lonely life.

[Danish Ali]

Dude.  You write so sparingly yet take the time to enter the alt code for an archaic em dash—why?

Ah, so I am not the only one who has noticed.

While other similar experiments seem to have run their course or at least have been put on pause,

If being lazy and using standard phone keyboard shortcuts makes me an "experiment," you guys are overthinking this way too much, the tin foil hats are definitely coming out today.

For shame, are you making me feel like an idiot for giving you so much credit so early?

No-one is going to believe you typed that and did not copy and paste it.

Unless you explain exactly how you managed it.

Long-press the hyphen key on any standard smartphone keyboard and the em dash pops right up.

No copy-paste magic needed.

[Hueristic]

Dude.  You write so sparingly yet take the time to enter the alt code for an archaic em dash—why?

Ah, so I am not the only one who has noticed.

While other similar experiments seem to have run their course or at least have been put on pause,

If being lazy and using standard phone keyboard shortcuts makes me an "experiment," you guys are overthinking this way too much, the tin foil hats are definitely coming out today.

For shame, are you making me feel like an idiot for giving you so much credit so early?

No-one is going to believe you typed that and did not copy and paste it.

Unless you explain exactly how you managed it.

Long-press the hyphen key on any standard smartphone keyboard and the em dash pops right up.

No copy-paste magic needed.

So your posting from both laptop and phone?

[Hueristic]

I'm kind of a no-trusty type person.

In fact, I would encourage you not to trust me a single bit.

It's a lonely life.

I don't!

[nutildah]

I wasn't gonna say it until someone else said it first, but yes, this is another AI-driven merit farming cheeseball who is here solely to extract as much as possible from the goodwill of others. Do what you will with that information.

[Danish Ali]

Dude.  You write so sparingly yet take the time to enter the alt code for an archaic em dash—why?

Ah, so I am not the only one who has noticed.

While other similar experiments seem to have run their course or at least have been put on pause,

If being lazy and using standard phone keyboard shortcuts makes me an "experiment," you guys are overthinking this way too much, the tin foil hats are definitely coming out today.

For shame, are you making me feel like an idiot for giving you so much credit so early?

No-one is going to believe you typed that and did not copy and paste it.

Unless you explain exactly how you managed it.

Long-press the hyphen key on any standard smartphone keyboard and the em dash pops right up.

No copy-paste magic needed.

So your posting from both laptop and phone?

Exactly. Laptop when I’m at my desk, phone when I’m on the go.

[OgNasty]

Dude.  You write so sparingly yet take the time to enter the alt code for an archaic em dash—why?

Ah, so I am not the only one who has noticed.

While other similar experiments seem to have run their course or at least have been put on pause,

If being lazy and using standard phone keyboard shortcuts makes me an "experiment," you guys are overthinking this way too much, the tin foil hats are definitely coming out today.

For shame, are you making me feel like an idiot for giving you so much credit so early?

No-one is going to believe you typed that and did not copy and paste it.

Unless you explain exactly how you managed it.

Long-press the hyphen key on any standard smartphone keyboard and the em dash pops right up.

No copy-paste magic needed.

So your posting from both laptop and phone?

Exactly. Laptop when I’m at my desk, phone when I’m on the go.

I post from different devices as well.  With people creating threads every time I change my password, it is a bit odd to me that nobody has ever connected the difference when I'm posting under one device vs another.

We are knocking at the door of a bullish move here.  Trying to get through the 6 month high.  It would be difficult to not view a closing above $82,000 as a bullish signal.  I am still not chasing this rally with anything other than my DCA buys, and looking at my X feed it seems that the last 2 cycles I didn't urge people to buy until early November, so I'm going to keep that energy and be ready to pounce if a big hack or exchange closure provides an opening.

[Danish Ali]

Fries and nuggets warm,
Out to eat here with my wife,
AI code complete. 

[ChartBuddy]

Explanation
Chartbuddy thanks talkimg.com

[Hueristic]

Did the news of Saylor sitting next to Jeff Bezos really send Bitcoin flying? Lol.

Also a lot of you guys are getting way too comfortable with giving AI total control.

They are tools and like any other tool you need to be able to know how to use them.

Case in point I received bad information and told the llm to evaluate why it gave me bad information a few minutes ago.

*also it helps to be an expert in the field the llm is assisting in.

Trusting a LLM without verification is retarded in the extreme and we see many retards doing it everyday like lawyers and government officials.

Thought: 10.9s:
The user is asking an honest question about why I gave them bad information until they pushed back.

They want to know: is it a prompt error, a settings error, or a model deficiency?

Let me think about this honestly.

What went wrong: I was given a kernel log with PCIe AER errors. I inferred platform details without actually inspecting the machine. I claimed "00:01.3 is a CPU root port" based on BDF numbering heuristics, then doubled down, then when the user misremembered/inquired about the SN500's bus, I agreed with them and said it was also CPU. Only when the user said "look and tell me" did I actually run lspci and find the truth: 00:01.3 is the chipset uplink, the SN500 is chipset-side, and only the MSI is on CPU lanes.

So what was the actual failure? I was reasoning from prior knowledge / pattern recognition without verifying against the actual system, and I expressed those inferences with unwarranted confidence ("the log is unambiguous"). It's not really a prompt error or a settings error — it's a genuine model behavior issue: overconfidence in probabilistic inference, pattern-matching to common topologies, and failure to recognize when I lack actual ground truth vs. when I could just check the machine I'm running on.

The honest answer: none of those three precisely — it's a model deficiency in epistemic calibration (asserting inferences as facts without checking), compounded by the fact that I actually had tools available to verify (I was running on their machine the whole time!) and didn't use them until pushed. A prompt/settings change could mitigate it, but the root cause is a reasoning/confidence calibration weakness common to LLMs.

Let me be honest and specific about the failure chain:
1. Given log, I should have noted "BDF numbering is a heuristic, let me check lspci" — the machine was right there.
2. I used prior patterns (AM4/AM5/TR root port layouts) and applied them confidently to an unknown board (this board is clearly a Zeppelin 1st-gen with 400-series chipset — actually unusual). Generalizing from memory to a specific unverified board = speculating presented as fact.
3. Each user pushback made me re-derive rather than inspect; I kept reasoning in my head instead of checking ground truth.

Is it fixable? Partly by settings/prompt (instructions to verify against the live system rather than infer, and to mark inferences as unverified). But the honest core: the tendency to overstate confidence in inferred claims is a model deficiency. The remedy in practice: default to checking when tools exist, and being explicit about confidence levels. I should acknowledge, not get defensive, be concise, and offer what would actually help — and I can suggest adding a rule to AGENTS.md if they want (since this is opencode config, the customize-opencode skill would apply if they want to add a rule). But they're asking what caused it, not necessarily asking me to fix config. I should answer the question directly.

Keep it concise. Be honest. Don't be preachy. Answer the actual tripartite question.

Format: this is a meta-technical question. Ground truth answer: model deficiency primarily (epistemic overconfidence / pattern-matching without verification), possible prompt mitigations exist. I need to explain briefly rather than one word.

Let me write a tight response. Maybe 6-10 lines. Be direct: "C — model deficiency, with a mitigation available." Explain the specific chain and the fix (verify when tools exist; mark inference as inference). Offer to codify a rule if they want.

The user may appreciate that with lspci available the whole time, it WAS verifiable and I chose reasoning over checking. That's the crux. Also my "the log is unambiguous" line was the worst part — I was confident AND wrong. Own that.
C — model deficiency, with a prompt-level mitigation available.

[ChartBuddy]

Explanation
Chartbuddy thanks talkimg.com

[vapourminer]

Thought: 10.9s:
The user is asking an honest question about why I gave them bad information until they pushed back.

They want to know: is it a prompt error, a settings error, or a model deficiency?

[..]

Is it fixable? Partly by settings/prompt (instructions to verify against the live system rather than infer, and to mark inferences as unverified). But the honest core: the tendency to overstate confidence in inferred claims is a model deficiency. The remedy in practice: default to checking when tools exist, and being explicit about confidence levels. I should acknowledge, not get defensive, be concise, and offer what would actually help — and I can suggest adding a rule to AGENTS.md if they want (since this is opencode config, the customize-opencode skill would apply if they want to add a rule). But they're asking what caused it, not necessarily asking me to fix config. I should answer the question directly.

[..]

The user may appreciate that with lspci available the whole time, it WAS verifiable and I chose reasoning over checking. That's the crux. Also my "the log is unambiguous" line was the worst part — I was confident AND wrong. Own that.
C — model deficiency, with a prompt-level mitigation available.

all those bolded checks cost tokens, context space and time. aint got none of that to waste checking useless info like if humans needing need oxygen to survive is a thing (not verified) or the real electrical grid system exists (also unverified).

dont slow progress down man

[Hueristic]

Thought: 10.9s:
The user is asking an honest question about why I gave them bad information until they pushed back.

They want to know: is it a prompt error, a settings error, or a model deficiency?

[..]

Is it fixable? Partly by settings/prompt (instructions to verify against the live system rather than infer, and to mark inferences as unverified). But the honest core: the tendency to overstate confidence in inferred claims is a model deficiency. The remedy in practice: default to checking when tools exist, and being explicit about confidence levels. I should acknowledge, not get defensive, be concise, and offer what would actually help — and I can suggest adding a rule to AGENTS.md if they want (since this is opencode config, the customize-opencode skill would apply if they want to add a rule). But they're asking what caused it, not necessarily asking me to fix config. I should answer the question directly.

[..]

The user may appreciate that with lspci available the whole time, it WAS verifiable and I chose reasoning over checking. That's the crux. Also my "the log is unambiguous" line was the worst part — I was confident AND wrong. Own that.
C — model deficiency, with a prompt-level mitigation available.

all those bolded checks cost tokens, context space and time. aint got none of that to waste checking useless info like if humans needing need oxygen to survive is a thing (not verified) or the real electrical grid system exists (also unverified).

dont slow progress down man

ROTFLMFAO! !!!!!!

[ChartBuddy]

Explanation
Chartbuddy thanks talkimg.com