This makes me recall a conversation from a podcast with Sam Harris where he discusses the “pornography of doubt”

Here is the YouTube clip, less than a minute long

https://youtube.com/shorts/ybUfy3DZK0U?si=o0t8AiLZE4XEeEYV

I use to subscribe to waking up app and really enjoyed the enlightening discussions with other intellectuals in different domains of biology, psychology and science.

The easiest way to understand why is by understanding natural language. A natural language like english is very messy and and doesn't follow formal rules. It's also not specific enough to provide instructions to a computer, that's why code was created.

The AI is incredibly dumb when it comes to complex tasks with long range contexts. It needs an engineer that understands how to write and execute code to give it precise instructions or it is useless.

Natural Language Processing is so complex, it started around the end of world war two and we are just now seeing innovation in AI where we can mimmick humans, where the AI can do certain things faster than humans. But thinking is not one of them.

It's like taking a student who wins a gold in IMO math, but can't solve easier math problems, because they did not study those type of problems. Where a human who is good at IMO math generalizes to all math problems.

It's just memorizing a trajectory as part of a specific goal. That's what RL is.

I've tried to think of specific follow-up questions that will help me understand your point of view, but other than "Cite some examples of easier problems than a successful IMO-level model will fail at," I've got nothing. Overfitting is always a risk, but if you can overfit to problems you haven't seen before, that's the fault of the test administrators for reusing old problem forms or otherwise not including enough variety.

GPT itself suggests[1] that problems involving heavy arithmetic would qualify, and I can see that being the case if the model isn't allowed to use tools. However, arithmetic doesn't require much in the way of reasoning, and in any case the best reasoning models are now quite decent at unaided arithmetic. Same for the tried-and-true 'strawberry' example GPT cites, involving introspection of its own tokens. Reasoning models are much better at that than base models. Unit conversions were another weakness in the past that no longer seems to crop up much.

So what would some present-day examples be, where models that can perform complex CoT tasks fail on simpler ones in ways that reveal that they aren't really "thinking?"

1: https://chatgpt.com/share/695be256-6024-800b-bbde-fd1a44f281...

“ This indicates that while CoT can improve performance on difficult questions, it can also introduce variability that causes errors on “easy” questions the model would otherwise answer correctly.”

Other response to strawberry example; There are 25,000 people employed globally that repair broken responses and create training data, a big whack-a-mole effort to remediate embarrassing errors.

For me the main lesson is, don't let your ego develop from success. Any human is vulnerable to narcissism. It is an interesting phenomenon, where you can originate as a humble person who becomes successful, only to lose your great qualities, when your identity changes. With success you attract different people in your life who may be attracted only to your success and who don't have the stones to confront you on your bs.

Developing healthy self awareness comes from surrounding yourself with people that love you, but are not afraid to keep you honest if you do something out of character.

All the exciting innovation over the past 13 years comes from deep learning mainly in working with images and natural language.

Machine learning is good for tabular data problems, particularly decision trees, that work well to reduce uncertainty for business outcomes, like sales and marketing as one example.

Machine Learning Basics:

Linear regression - Y = Mx + B (predicts a future value) Classification (logistic regression) - Y = 1 / 1 + e^-(b0 + b1x) (predicts probability of a class or future event)

There is a common learning process between the two called gradient descent. It starts with the loss function, that measures the error between predictions and ground truth, where you backpropogate the errors as a feedback signal to update the learned weights which are the parameters of your ml model which is a more meaningful representation of your dataset that you train on.

In deep learning it's more appropriate for perception problems, like vision ,language and time sequences. It gets more complex where you are dealing with significantly more parameters in the millions, that are organized in hierarchical layer representation.

There are different layers for different types of learning representation, Convolutions for Images and RNN for Sequence to Sequence learning and many more examples of layers, which are the basis of all deep learning models.

So there is a small conceptual overlap; but I would say deep learning has a wider variety of interesting applications, is much more challenging to learn, but not impossible by any stretch.

There is no harm in giving it a try and diving in. If you get lost and drown in complexity, start with machine learning. It took me 3 years to grasp, so it's a marathon, not a sprint.

Hope this helps

In regards to deep learning, building deep learning architecture is one of my greatest joys in finding insights from perceptual data. Right now, I'm working on spatiotemporal data modeling to build prediction systems for urban planning to improve public transportation systems. I build ML infrastructure too and plan to release an app that deploys the model in the wild within event streams of transit systems.

It took me a month to master the basics and I've spent a lot of time with online learning, with Deeplearning.ai and skills.google. Deeplearning.ai is ok, but I felt the concepts a bit dated. The ML path at skills.google is excellent and gives a practical understanding of ML infrastructure, optimization and how to work with gpus and tpus (15x faster than gpus).

But the best source of learning for me personally and makes me a confident practitioner is the book by Francois Chollet, the creator of Keras. His book, "Deep Learning with Python", really removed any ambiguity I've had about deep learning and AI in general. Francois is extremely generous in how he explains how deep learning works, over the backdrop of 70 years of deep learning research. Francois keeps it updated and the third revision was made in September 2025 - its available online for free if you don't want to pay for it. He gives you the recipe for building a GPT and Diffusion models, but starts from the ground floor basics of tensor operations and computation graphs. I would go through it again from start to finish, it is so well written and enjoyable to follow.

The most important lesson he discusses is that "Deep learning is more of an art than a science". To get something working takes a good amount of practice and the results on how things work can't always be explained.

He includes notebooks with detailed code examples with Tensorflow, Pytorch and Jax as back ends.

Deep learning is a great skill to have. After reading this book, I can recreate scientific abstracts and deploy the models into production systems. I am very grateful to have these skills and I encourage anyone with deep curiosity like me to go all in on deep learning.

I’m curious how ML/AI is leveraged in the domain of public transport. And what can it offer when compared to agent based models.

Here is a scientific abstract that inspired my to start building this system. -> https://arxiv.org/html/2510.03121

I am unfamiliar with agent based models, sorry I can’t offer any personal insight there, but I ran your question through Gemini and here is the AI response:

Based on the scientific abstract of the paper *"Real Time Headway Predictions in Urban Rail Systems and Implications for Service Control: A Deep Learning Approach"* (arXiv:2510.03121), agent-based models (ABMs) and deep learning (DL) approaches compare as follows:

### 1. Computational Efficiency and Real-Time Application

* *Deep Learning (DL):* The paper proposes a *ConvLSTM* (Convolutional Long Short-Term Memory) framework designed for high computational efficiency. It is specifically intended to provide real-time predictions, enabling dispatchers to evaluate operational decisions instantly. * *Agent-Based Models (ABM):* While the paper does not use ABMs, it contrasts its DL approach with traditional *"computationally intensive simulations"*—a category that includes microscopic agent-based models. ABMs often require significant processing time to simulate individual train and passenger interactions, making them less suitable for immediate, real-time dispatching decisions during operations.

### 2. Modeling Methodology

* *Deep Learning (DL):* The approach is *data-driven*, learning spatiotemporal patterns and the propagation of train headways from historical datasets. It captures spatial dependencies (between stations) and temporal evolution (over time) through convolutional filters and memory states without needing explicit rules for train behavior. * *Agent-Based Models (ABM):* These are typically *rule-based and bottom-up*, modeling the movement of each train "agent" based on signaling rules, spacing, and train-following logic. While highly detailed, they require precise calibration of individual agent parameters.

### 3. Handling Operational Control

* *Deep Learning (DL):* A key innovation in this paper is the direct integration of *target terminal headways* (dispatcher decisions) as inputs. This allows the model to predict the downstream impacts of a specific control action (like holding a train) by processing it as a data feature. * *Agent-Based Models (ABM):* To evaluate a dispatcher's decision in an ABM, the entire simulation must typically be re-run with new parameters for the affected agents, which is time-consuming and difficult to scale across an entire metro line in real-time.

### 4. Use Case Scenarios

* *Deep Learning (DL):* Optimized for *proactive operational control* and real-time decision-making. It is most effective when large amounts of historical tracking data are available to train the spatiotemporal relationships. * *Agent-Based Models (ABM):* Often preferred for *off-line evaluation* of complex infrastructure changes, bottleneck mitigation strategies, or microscopic safety analyses where the "why" behind individual train behavior is more important than prediction speed.

I love hn and was inspired by all the devs who have their own site. I was drowning in work, but put the Django architecture together on vacation, started putting things together today and it’s been a blast.

I don’t enjoy social media and was thinking to posse intrinsically.

I appreciate this post and the authors perspective.

If you're running a server anyway, it seems trivial to serve content dynamically generated from markdown - all an SSG pipeline adds is more dependencies and stuff to break.

I know there's a fair few big nerd blogs powered by static sites, but when you really consider the full stack and frequency of work that's being done or the number of 3rd party external services they're having to depend on, they'd have been better by many metrics if the nerds had just written themselves a custom backend from the start.

It could easily have been a static website, but I happened to stumble across PWS, which came bundled with a default ASP website. That is how I got started. I replaced the default index.asp with my own and began building from there. A nice bonus of this approach was that the default website included a server-side guestbook application that stored comments in an MS Access database. Reading through its source code taught me server-side scripting. I used that newfound knowledge to write my own server-side applications.

Of course, this was a long time ago. That website still exists but today most of it is just a collection of static HTML files generated by a Common Lisp program I wrote for myself. The only parts that are not static are the guestbook and comment forms, which are implemented in CL using Hunchentoot.

I probably wouldn't be able to handle 0.5M requests, but I am nowhere near getting them. If I start approaching such numbers I'll consider an upgrade.

Check out Wagtail if you'd like to have even more batteries included for your site, it was a delight building my site with it:

https://blog.miloslavhomer.cz/hello-wagtail/

I'd still recommend starting with SQLite, seems that by skipping a DB service you can save quite a few bucks.

>> I think I’m particularly suspicious of community, because as a writer and pedantic arsehole on the internet, I value truth-seeking behaviour. I want people to think and say things that are true, not just things that they have to believe for the sake of keeping their community happy.

Unfortunately, this is what happens with every group of people.

Our individual realities are highly subjective. A group of people who are part of a community construct a shared reality that they can all accept. If you don’t contribute to the shared reality, you are treated as someone who is problematic.

As humans we are social creatures. In our evolution, we develop cognitive systems that help us thrive in social structures. One system is called the social protection system. This system gets activated when we sense tension in relationships and sends a signal of fear to the subject that they risk being separated from a social group. This fear motivates people to maintain connection. So some people are intrinsically motivated by fear to maintain their status, sometimes unconsciously.

Our self esteem comes from two things, relationships and mastery. Healthy self esteem comes from connection to people who accept you for who you are, where you feel visible and accepted with your good and bad traits.

If you have a few people in your life with this type of connection, you will have a healthy social foundation and rely less on belonging to a group.

Groups are valuable in that the human experience is complicated. The best source of information comes directly from other humans and their experiences overcoming complexity.

However, I do agree with the author where certain groups can be problematic, particularly engaging in things like tribalism.

Establishing good self esteem by keeping a few people close to you who see you and accept you as a flawed human is key. The other part is to immerse yourself in activities where you develop mastery and maintain a connection to the activities that are intrinsically motivating and satisfying without distraction from external signals.

I learned this by studying the science of self actualization, from the research done by Scott Barry Kaufman and his book Transcend. He’s a humanistic psychologist who was inspired by Abraham Maslow, one of the founders of humanistic psychology.

* His shareholders include many Americans. The cash pile is concentrated as far as corporate governance is concerned, but lots of people own a piece of it in their retirement accounts. (Often indirectly via index funds.)

* Many employees work for businesses owned by Berkshire Hathaway. I'd guess they're pretty stable businesses, not about to get bought out and shut down? Maybe decent places to work? (Not sure about the railroad, though.)

* Many people are customers of these businesses. Mostly a good thing?

Admittedly there are lots of Americans that don't participate in that, though.

Finding healthy ways to blow off steam maintains continuity with your most important goals.

Well done!