HNHacker News
TopNewBestAskShowJobs

sgt101

7,940 karma · joined January 24, 2012

nothing to see here, move along, move along.
submissionscomments
sgt101··on Apprentice, Journeyman, and Master: The Medieval Guild (2018)
This structure is good for engagements that have long duration (like >3mths) and high intensity.. a lead, an heir and a spare covering a key "role" means that you have continuity if someone goes sick or gets another job. The project can just keep rolling along!
sgt101··on The reanimation of pseudoscience in machine learning
These are examples and the experimental designs were a particular choice by the authors of those examples - they aren't intrinsic to ML or the ML community.
sgt101··on The reanimation of pseudoscience in machine learning
To use the authors own framework the epistemic failure of this paper is the mistake of identifying the failures of some individuals with a failure of an institution or a technology. The fact that harm can be created by some individuals means nothing more than that. There are pseudoscientists in ML, but to generalise that fact into an argument that there is something rotten at the heart of ML is just foolish. The authors also fail to understand that the idea of a causal theory as fundamental to science is extremely tenuous and has really only applied to physics for most of the history of science. The workings of plants (for example) were understood with an almost complete ignorance of the mechanisms that caused them to behave in particular ways until relatively recently. This didn't impair the value of this observational and contingent knowledge. The comprehension of gravity as a field determining the inverse square law was so brilliant and important that it's blinded us to the reality of so many other fields of knowledge - and their legitimacy.
sgt101··on Cargo Cult Science (1974)
I think that there is a pernicious impact on many folks because they do "all the right things" and still struggle mightily with their work and responsibilities because doing 80 hard hours of the wrong work doesn't deliver.
sgt101··on Creativity fundamentally comes from memorization?
This is better : https://www.interaliamag.org/articles/margaret-boden-creativ...
sgt101··on Small neural network enables realistic rendering of woven fabrics in real-time
Resnet + 16 layers isn't GPT4 but it's not small.

I think that the 60fps on a consumer grade card is the win, but that says a lot about the power of consumer grade cards in 2024.

sgt101··on Bayesian Statistics: The three cultures
Bayes lets you use your priors, which can be very helpful.

I got all riled up when I saw you wrote "correct", I can't really explain why... but I just feel that we need to keep an open mind. These approaches to data are choices at the end of the day... Was Einstein a Bayesian? (spoiler: no)

sgt101··on Crash course in deep learning for computer graphics
What has this got to do with computer graphics?
sgt101··on Taking my diabetes treatment into my own hands
It depends on genetics and luck.

The luck part is that it seems that infections trigger the autoimmune reaction that kills the pancreas. The genetics bit is that you may or may not have got the gene from your father.

Most people die of heart attacks in the end. Factors like smoking, lifestyle and fighting in wars are probably more important than well managed type 1 nowadays. The big difference now is that the insulin is human insulin, made by genetically engineered microbes. In the past it was harvested from animals and it didn't work as well. Also constant blood monitoring means that highs and lows can be detected and fixed before damage is done. So - things have moved on, there isn't as much to be frightened of, I'm sorry your dad died young, but you will probably be ok.

sgt101··on Taking my diabetes treatment into my own hands
T1 needs insulin, some T2 are ok without it.

Insulin is cheap to make, now, it is expensive because of commercial considerations like monopolization or investment. In reality any national system worth its salt could produce enough insulin at a very low cost for all diabetics in the world. But, this won't happen because of trade rules and so on.

Some people are trying to build the infrastructure for local/homebrew insulin production, but it's proving to be challenging. See this site for more: https://openinsulin.org/2023-recap/

sgt101··on Taking my diabetes treatment into my own hands
What is it that Camaps + dexcom doesn't do that you want?
sgt101··on Taking my diabetes treatment into my own hands
This is counter to my experience - my daughter has received fantastic care. We have regular time with the endocrinologist and get phoned up inbetween clinics. They have provided a closed loop system and all the backup we could have asked for.

I agree about Glooko, it's not as good as diasend was.

sgt101··on Can the New Mathstral LLM Accurately Compare 9.11 and 9.9?
you are not confused. >* we can conclude that:\n\n$$9.11 > 9.9$$\n\n7.* Therefore, the final answer is:\n\n$$\\boxed{9.11}$$","tool_calls":[],"role":"assistant"},"finish_reason":"stop","logprobs":null}],"usage":{"prompt_tokens":26,"completion_tokens":293,"total_tokens":319}}
sgt101··on Money and Happiness: Extended Evidence Against Satiation
this seems to be "people who have to work are less happy than people who are free to do what ever they want whenever they want"
sgt101··on We need visual programming. No, not like that
People did some work on this : https://www.google.co.uk/books/edition/Drawing_Programs_The_...
sgt101··on Europe's new heavy-lift rocket, Ariane 6, made its inaugural flight
Yes, and the industrial base that we are discussing is the one that builds the M51.3 which is fundamental for Europe. Existence is worth a lot.
sgt101··on Introduction to Program Synthesis
Thanks - this is extremely insightful and helpful.
sgt101··on Introduction to Program Synthesis
This is super interesting, thank you.

But, looking at semdesign's site and info I have a question - which is if this tech exists since 2010 or so... why are all the big banks not using it? Is it just bad marketing or is there a gotcha? They are all whining about their cobol problem, this could be a fix......

Does this mean that the market is not what folks imagine?

sgt101··on Introduction to Program Synthesis
OK!

I work in financial services, a lot of customers want tools to support their creaking code bases.

What are the state of the art options, both commercial and open source?

What are the prospects for things like translating COBOL to Python or Java?

sgt101··on Python with Braces
I don't like whitespace compulsion but I can live with it.

self.onandonandon is a real pain along with copy.deepcopy

sgt101··on Our great database migration
utterly utterly unhinged.

It's basically a design that someone both very naive and suffering late stage Rabies would come up with.

sgt101··on Our great database migration
I depends what you are putting into it... I think it's completely feasible to get 100000 customer records into 80 MB... and then some
sgt101··on Our great database migration
You. Were. Running. An. Insurance. Company. On. SQLite?

What?

What possessed them?

sgt101··on Vigorous Exercise, Cognitive Decline, and High Blood Pressure
I have only had a quick look, but I think that this is suss because of the VPA definition. The authors could show us the data vs. response to the questionnaire, in particular the columns for folks doing >5 episodes of high intensity exercise per week... but they don't and instead simply show the categories >1per week and <1 per week.

This smells bad.

sgt101··on How to run an LLM on your PC, not in the cloud, in less than 10 minutes
Any M-series mac with 16GB+ can do this also.
sgt101··on Babylon's Mystery Goddess
Wikipedia is good on the artefact : https://en.wikipedia.org/wiki/Burney_Relief

This includes the reconstruction of the original colours: https://en.wikipedia.org/wiki/Burney_Relief#/media/File:Quee...

sgt101··on Nature retracts paper that claimed adult stem cell could become any type of cell
>>It's not worth it (or rather: extremely high financial risk) for companies or individuals to do it

The problem is that it's hard to reliably capture value from research. A good example is LLM's. If OpenAI, Google, Meta and AWS had been able to build a wall round GTP3.5 Turbo and above models then I expect that they could have captured all the value of the research effort.. as it is I don't think that is/will be the case - it's almost too easy to replicate as Mistral have shown. Note: I'm not saying it's trivial or something, but if you spend a few $million on it you can get close enough, and then spending a few $million more will get you all the way. Also, I am not talking about building a frontier model today (which requires $100millon or so and some difficult skills/organisation) but rather a model in say 3 years time with the frontier performance of todays models.

sgt101··on Creativity has left the chat: The price of debiasing language models
No doubt you could construct a more diverse set, but it still wouldn't be creative.
sgt101··on Creativity has left the chat: The price of debiasing language models
No, diversity isn't creativity. For example, we could search google for "great art" and if it produced a sample of one art work from ever decade of the last 500 years that would likely be highly diverse in style and content. If it returned a list of the best work from western Europe in the of the 18th century it would be rather consistent. Both lists would have the same amount of creativity though - 0.
sgt101··on Creativity has left the chat: The price of debiasing language models
I wish that the author hadn't described semantic and syntactic diversity as creativity.
← PreviousPage 16 of 34Next →