HNHacker News
TopNewBestAskShowJobs

sgt101

7,940 karma · joined January 24, 2012

nothing to see here, move along, move along.
submissionscomments
sgt101··on Context engineering
Why would I believe that any of this works? This is just some blokes idea of what people should do.

There is no evidence offered. No attempt to measure the benefits.

sgt101··on Amazon confirms 14,000 job losses in corporate division
How much of the search market will Agents or Chatbots take?

Google clearly believe : a chunk.

People switched to GPT and Claude...

Amazon risk becoming logistics and datacenters, what's wrong with that you ask? Well, it's not where the money is. The money is in the website and the application layer.

sgt101··on Amazon confirms 14,000 job losses in corporate division
They had the infrastructure and muscle to put a dent into Google Search and Microsoft Enterprise software... but instead they've ceeded to OpenAI and Anthropic.

They had a bridgehead device of the type that OpenAI are struggling to create with Jonny Ive in Alexa... but nothing seems to be happening with Alexa now.

sgt101··on Claudeskills.cc – Share, Discover, and Reuse Claude/OpenAI Agent Skills
Do you have any evidence that this works? Can you link the evaluation and result?
sgt101··on A small number of samples can poison LLMs of any size
This is harsh on HN
sgt101··on Dutch government takes control of Chinese-owned chipmaker Nexperia
1. OP and me were talking about the US government.

2. It's quite possible you are right about Chinese EUV capabilities - I don't have a clue myself as I don't work for NSA or CIA. On the other hand it's one more very big challenge in a field of big challenges and despite what the CCP says Chinese resources are very far from infinite and the clock is ticking demographically.

3. A set of EUV test runs would not impress me to be honest - a full blown production line of EUV generated chips would. The issue doesn't seem to be in doing EUV, the issue is to build systems that can sustain quality at volume.

sgt101··on Dutch government takes control of Chinese-owned chipmaker Nexperia
Where else will they buy EUV from?
sgt101··on John Searle has died
>translational states to be strong enclitic feature of various judgements of the intermediary

I don't understand, could you explain what you mean?

I looked up enclitic - it seems to mean the shortening of a word by emphasizing another word, I can't understand why this would apply to the judgements of an intermediary

sgt101··on John Searle has died
I have always wondered if we would be capable of writing down the mechanisms that power our thoughts. I think that this was one of the ideas that bubbled up from reading Godel Escher Bach many years ago. Is it possible for us to express the machine that makes us using the outputs of that machine in the way that it's not possible to write second order logic using first order logic.

Of course, also there are processes that are not expressible as computations, but those of these that I know about seem very very distant from human thought, and it seems very very improbable that they could be implemented with a brain. I also think that these are not observed in our universe so far.

sgt101··on John Searle has died
>also drawing on layers of abstractions where it is easy to mistakenly attribute features of these abstractions to reality.

Ok - I get that bit. I have always thought that physics is a description of the universe as observed and of course the description could be misleading in some way.

>the methods of physics intentionally ignore certain aspects of reality and focus on quantifiable and structural aspects

Can you share the aspects of reality that physics ignores? What parts of reality are unquantifiable and not structural?

sgt101··on John Searle has died
>Understanding / cognition is not necessarily physical and certainly not well defined.

Whooha! If it's not physical what is it? How does something that's not physical interact with the universe and how does the universe interact with it? Where does the energy come from and go? Why would that process not be a physical process like any other?

sgt101··on A small number of samples can poison LLMs of any size
Another point = we can inspect the contents of the wikipedia page, and potentially correct it, we (as users) cannot determine why an LLM is outputting a something, or what the basis of that assertion is, and we cannot correct it.
sgt101··on Vibe engineering
So what you're telling me is that I am genetically predisposed to brew and drink beer?

Explains a lot.

Explains it to my satisfaction!

sgt101··on Vibe engineering
Although, gradually, all over the world people chose that lifestyle rather than take their chances with the seasons and the hunt.
sgt101··on Vibe engineering
Maybe it's in my head, but I have completed coding projects that I believe would have taken a team of five offshore maybe 12 weeks to do in the past in about ten working days while juggling calls and living normal corporate life.

The win is that I don't have to share the vision of what needs to be done and how it should all work, and then constantly monitor and reframe that in the face of the teams missteps and real findings. I work with the agents directly, and provided I set the architecture and build up systematically I can get really good results. The cycle time between me identifying an issue and the issue getting fixed by me and the agents is now minutes rather than hours or days with an off shore team. Even better the agents can provide bug fixing expertise much quicker than stack overflow - so I can figure out what's wrong much faster so as to specific what needs fixing.

It is no good walking in and requesting functionality, you need to know how the thing you want should work, and you need to know what good looks like, and what bad looks like, and how good is separated from bad. Then the normal process of discovery ("eep that doesn't actually work like I thought") can take place and you can refactor and repair as required.

Sometimes I start something that just doesn't work, you have to recognise that you and the agents are lost, and everything needs to be torn down. You then need to think properly about whats gone wrong and why, and then come back with a better idea. Again - just like with dev teams, but much more clearly and much faster.

sgt101··on LoRA Without Regret
I just think that there has to be some heuristic..
sgt101··on LoRA Without Regret
Question for dudes building modern nn's... what's the thinking on estimating structural capacity for real world problem? How should I estimate how many parameters to choose for the model?
sgt101··on Leaked Apple M5 9 core Geekbench scores
I had a 16' m1 pro until July when work had an M3 with 2GB more ram going free and I thought it would be smart to take it, in the past when I had had a three+ year old machine and upgraded everything was super better.. but this time I don't notice any difference at all in almost any application.

The only place I feel it is when I am running a local llm - I do get appreciably more tokens per second.

sgt101··on HSBC Claims Quantum Trading Breakthrough
I am on really shaky ground, but I think that they are representing "uncertainty" in the QC sense per event and that they claim that over the totality of the events this then represents reality better than stochastic sampling.

But I've been reading the paper on and off today and I don't really understand much of it at all.

sgt101··on HSBC Claims Quantum Trading Breakthrough
https://arxiv.org/pdf/2509.17715 for the paper.

- this is back testing. - the mechanism (from first skim) is to create features using the QC which are then fed to the classical algorithm.

sgt101··on The Data Commons Model Context Protocol (MCP) Server
ok I know that this is MCP but basically it's a query...
sgt101··on The Data Commons Model Context Protocol (MCP) Server
I like the idea of this but I worry about the applications.

Am I alone in reading "Agents, however, understand complex queries and are able to fetch and compile the needed data quickly." and thinking "bullshit". The fact is that LLM data discover/assistant apps that I have built with the current frontier models do a pretty good job of transforming language to sql... but do cock things up fairly often. If you present the query to the user and do a fair amount of cross checking you can create a decent app... But the language of "all your problems are over" is just misleading.

sgt101··on UK Millionaire exodus did not occur, study reveals
>That isn't the case. Aging has very little to do with our issues. The amount of revenue generated by people in their 30s is nowhere near enough to pay for the state as it is. The biggest marginal increase in sickness benefits has been people in 20/30s.

Apart from "That isn't the case" (great point there) I don't understand how any part of this paragraph refutes what I wrote.

But hey, belief beats stats every time I guess.

sgt101··on UK Millionaire exodus did not occur, study reveals
My analysis is that if 55-64 year old were as economically active as 45-54 year olds the deficit would disappear.

The way to get this to happen is to stamp on age discrimination, provide more retraining, improve pension incentives, and tightening sick benefits sharply.

sgt101··on UK Millionaire exodus did not occur, study reveals
But whatever happens the uk will not lose tax income from these people. Yes, they will not buy expensive property. Yes they will not buy designer goods... but if they leave will the impact on the average citizen be at all significant?
sgt101··on Bertrand Russell to Oswald Mosley (1962)
And yet here we are...
sgt101··on UK launches Project Octopus to deliver interceptor drones to Ukraine
I think Anduril not Palantir?
sgt101··on UK launches Project Octopus to deliver interceptor drones to Ukraine
Size and range.

Strike drones have to be able to carry a fairly large warhead (or are only good at hitting people and not things) and they have to fly quite a long way to get at things like reserve assets and logistics. So they are quite big, with quite a lot of fuel etc. Big things tend to cost more. In this case I can imagine that an interceptor that has a range of 10k and is 5% of the size of the strike drone would be able to knock it down and would be able to do so well away from its target.

Dunno how anyone can "know" unless they "know" and then they are not talking. But, it seems plausible that something with 10% of the range and 5% of the mass would cost 10% or less.

sgt101··on UK launches Project Octopus to deliver interceptor drones to Ukraine
One approach is directed energy, there are laser guns like the UK's dragonfire (there are many others out there too) however these have problems in dusty or foggy conditions for obvious reasons. There are also microwave effectors which are used to fry the electronics on drones. These take advantage of the advances in Gallium Nitride based power electronics (and other even more exotic materials).
sgt101··on Learning lessons from the loss of the Norwegian frigate Helge Ingstad
Compare and contrast to the recent collision of Chinese combat ships around the Philippines - much more severe damage, much less severe outcome.

Someone needs to take a very hard look at this.

← PreviousPage 5 of 34Next →