View allAll Photos Tagged debuging
no HUD, debug camera. Sometimes having to get far enough away to hide the dataverse overlays produces a good angle.
Don't watch if spinning makes you feel bad. I am still working on my Flatland program (to find a bug that occurs when new earthquakes are plotted, but only sometimes) and found some debug tools I added to the code years ago. The one shown in the video spins the camera (making the entire scene spin about) for some reason. I thought it was a very silly thing for me to add...
DeBug Computer Nevada Day Parade 2006
Visit us on the web www.debugcomputer.net/?utm_source=flickr&utm_medium=p...
@aaronpk debugging message send issues at Jam on Hawthorne that caused a break in filming a demo video.
6 Likes on Instagram
4 Comments on Instagram:
car_o_lyn_brat_no_ber: that is so my jam!
kodinlanewave: Is it me or does he look like obi wan in this shot?
caseorganic: @kodinlanewave Yes he does!
caseorganic: @car_o_lyn_brat_no_ber Come back to visit!
DeBug Computer in the 2008 Nevada Day Parade
Visit us on the web www.debugcomputer.net/?utm_source=flickr&utm_medium=p...
DeBug Computer in the 2009 Nevada Day Parade
Visit us on the web www.debugcomputer.net/?utm_source=flickr&utm_medium=p...
DeBug Computer in the 2008 Nevada Day Parade
Visit us on the web www.debugcomputer.net/?utm_source=flickr&utm_medium=p...
Who's safe kiddos? Glad I don't make a living in Information Technology or work in an office doing any manual input... Interesting now with the way we are going with AI. "LLM" Large Language models will probably wipe out a lot of Data Entry Clerks and Administrators. they are highly susceptible to automation due to repetitive, structured, and unstructured data handling: Customer Service Representatives & Support: Chatbots and AI assistants/agents are managing routine customer inquiries and internal queries. Junior Software Developers & Coders: AI-powered tools are now able to write, debug, and test code, reducing the need for entry-level human developers.
Quality Assurance (QA) Testers: Automated testing tools are becoming more intelligent and faster than manual testers.
Graphic Designers & Content Writers: Generative AI tools (e.g., DALL-E, GPT) can rapidly produce content and designs, lowering demand for entry-level creative work.
Translators and Interpreters: High-level AI language models have rendered traditional translation work largely obsolete.
IT Support/System Administrators: Routine troubleshooting is increasingly handled by AI diagnostics.
I was contracted in the early 2000's by Dell Computers through "Tecnet" --my employer at the time. Dell was based out of Texas and if you reached the call center, there was a person with a strong Texas drawl - similar to the traditional "Deep South" drawl- hard to understand. This soon changed to cost saving remote call centers in India. This arrangement didn't last long at first because people, especially the Texans couldn't understand the East Asian accent. Funny to think but a lot of us couldn't understand either of these parties but the Texans talking to the East Asians is funny to ponder. Call Centers went back to Texas temporarily but then returned to India shortly after. I remember reading somewhere they had training programs for people to lose their accent in Asia call centers.
BB and Reddit: My searches 2026. India Built the World’s Back Offices in mass. A.I. Is Starting to Shrink It. "Artificial intelligence promises to automate the white-collar work that made India a tech powerhouse. The country is racing to adapt before it’s too late."
BB--I'm trying to understand AI myself and help others in laymen's terms!
Linguistic Intelligence is really just the beginning of AI.
What is the Socratic Method of teaching?
Instead of giving information and facts, an instructor using the Socratic method of teaching asks students a series of open-ended questions (questions with more than a yes or no answer) about a specific topic or issue. In turn, the students can also pose questions of their own.
So , I will include the betterment Ideas of LLM instead of the best word inference in the chain that is typical of LLM--
LLM's are so much better when instructed to be socratic.
This idea basically started from Grok, but it has been extremely efficient when used in other models as well, for example in Google's Gemini.
Sometimes it actually leads to a better and deeper understanding of the subject you're discussing about, thus forcing you to think instead to just consume its output.
It works with some simple instructions saved in Gemini's memory. It may feel boring at first, but it will be worth it at the end of the conversation.
Of course AIs do not have critical thinking, but they can trigger and impel their user to activate their own critical thinking capacity. That is the value of a Socratic exchange, it aids the human user in the use of their own, perhaps weak, critical analysis. Over time and repeated use of Socratic exchanges with an LLM, a user will develop greater, better than before, cognitive abilities. I'm not keeping the references to the proof, but this has been demonstrated and published in Psychology journals.
When you tell an LLM to be Socratic, you aren’t magically making it “smarter.” What you’re really doing is reorganizing the interaction loop. Rather than the model collapsing uncertainty into one elegant, finalized response, you’re prompting it to keep the reasoning space open longer. That alters the nature of the conversation.
For example, if you ask Why do startups fail?
a default response might give you a clean list: poor product-market fit, funding issues, bad leadership, etc. It feels complete. But if the model is instructed to be Socratic, it might respond with: Are you asking from the perspective of a founder, investor, or policymaker? or Are you more interested in early-stage failure or scale-stage collapse?
Suddenly, the reasoning space widens before it narrows. The discussion becomes shaped rather than delivered.
LLMs are essentially next-token predictors trained on patterns of conversation and exposition.
By default, they optimize for completion ..they produce something coherent and finished. In Socratic instruction, the objective shifts from answer production to guided exploration. And that shift alone often increases engagement. Consider a student asking, “What is justice?”
A standard response might summarize Rawls, Aristotle, and utilitarianism in a neat paragraph.
A Socratic version might ask: Do you think justice is about fairness, equality, or desert? and Can a system ever produce unequal outcomes?
Now the student has to think. The model hasn’t just transferred information, but it has activated cognition.
Here’s the additional perspective: It’s not only about clearer understanding for the user but also about distributed cognition between human and model.
When the model asks questions back, it externalizes intermediate reasoning steps that would otherwise remain compressed. In a typical answer, much of the reasoning is hidden behind the final synthesis.
In a Socratic exchange, those intermediate steps become interactive checkpoints.
Take a practical case: User: How do I improve my productivity? Default model: gives 10 tips.
Socratic model: What currently distracts you most is digital interruptions, unclear goals, or energy levels?
Now the human provides constraints. The model adapts. The final strategy emerges collaboratively. The intelligence is co-constructed rather than pre-packaged.
So the gain is not merely a feature of the model, it’s a feature of the interaction protocol.
There’s also a cognitive forcing function at work. When models ask clarifying questions, they narrow the hypothesis space and reduce hallucination risk. Instead of guessing what the user means, they query ambiguity directly.
For instance, if a user asks, “Explain the impact of the revolution,” that’s dangerously underspecified. Which revolution? French? Industrial? Digital? A default answer risks misalignment.
A Socratic response might begin: “Which revolution are you referring to, and in what context — political, economic, or technological?” That clarification increases epistemic alignment before any claim is made.
However, there is a tradeoff. Socratic prompting increases depth but reduces throughput. It is inefficient if the task is quick synthesis. If you ask, “What’s the capital of Japan?” a Socratic reply asking, “Are you preparing for a geography exam or planning travel?” is unnecessary friction.
It shines when the task involves: Conceptual learning (e.g., understanding entropy beyond a definition) Moral or philosophical inquiry (e.g., debating free will) Ambiguous problem framing (e.g., defining strategy before execution) Creative exploration (e.g., shaping a novel’s theme through iterative refinement) It is less useful for: Factual lookups Structured output tasks (e.g., “Format this as JSON”) Deterministic problem-solving (e.g., “Solve this equation”)
Socratic prompting does not universally enhance LLM performance. It restructures the reasoning topology of the exchange. It shifts the model from an answer engine to a cognitive scaffold. And perhaps the deeper insight is this: as LLMs grow more capable, the limiting factor increasingly becomes question quality rather than raw model intelligence.
For example, two users can query the same powerful model.
User A asks: Tell me about economics. User B, guided Socratically, refines through dialogue: I’m trying to understand why inflation hurts borrowers differently than lenders — can we unpack that step by step?
The second interaction produces deeper understanding not because the model changed, but because the questioning improved. A Socratic mode doesn’t merely enhance outputs. It upgrades the human participant in the loop. That is why it feels more powerful.
DeBug Computer Nevada Day Parade 2006
Visit us on the web www.debugcomputer.net/?utm_source=flickr&utm_medium=p...
DeBug Computer in the 2009 Nevada Day Parade
Visit us on the web www.debugcomputer.net/?utm_source=flickr&utm_medium=p...
have a public YT [YouTube] video:
www.youtube.com/watch?v=sMfNwA-hT7Q
that plays nicely
added the same to a YTM [YouTube Music] playlist:
but it's not playing through the YTM playlist
where's the setting to enable that a YT video should play under YTM‽
~~~~~
found an interesting thing during the #debug process:
replace the www in the YT link with music to play it in YTM
The YT video plays in YTM with:
music.youtube.com/watch?v=sMfNwA-hT7Q
am able to add it to the Music for Monday playlist
but it shows up there with a !, indicating that it can't be played
~~~~~
some catch is there somewhere that we aren't able to crack