teruakohatu

> This is just PR and carbon greenwashing.

> All current power plants are completely unsuitable as single power source for a data center

I don’t think this is a fair assessment.

For example, there is a data center being built in New Zealand that will be grid connected but the power will be supplied from a huge hydro dam that has to have generators shutdown if the load is not great enough. Its primarily purpose is to power an aluminium smelter and there is not enough transmission capacity to transfer all the electricity elsewhere.

Another counter example is home solar. If a house is grid connected it’s still producing green energy even if it still sips gas generation at night.

> But paradoxically there is simultaneously a lack of top tier within New Zealand

> Really? I'm aware of some extremely top-tier and wildly underemployed talent

In my experience it is incredibly hard to hire top-tier talent, but both of our experiences could simultaneously be true.

Are these people you know actively applying for better jobs and not getting them?

I know of three excellent devs who are IMHO vastly unemployed. At least two of them would struggle to get through a corperate interview process. The other is happy with the chill job they have.

I can confirm this is the case when my organisation advertises roles they get inundated with applications.

But paradoxically there is simultaneously a lack of top tier within New Zealand.

I the past year I have turned down an unsolicited job offer and I am also aware of two or three roles in two organisations that are not being advertised for due to lack of available talent.

New Zealand has a *lot* of potential, but this potential eventually ends up in Australia, London or the USA.

Countries should aspired to be an anode, not a cathode.

The politicians [1] I have spoken to about this issue generally don't consider it to be a pressing matter and are happy for New Zealand firms to move the HQ overseas if they are employing kiwis locally, in much the same way US firms might open an office in Asia to take advantage of lower wages.

[1] One nice thing about New Zealand is that you can get face to face time with a Member of Parliament easily, and Cabinet Minister if you are persistent.

Have you don’t any transmitting with the HackRF One? What was the experience like?

I guess it depends if you want to restrict an agent to a set of protocols or let it go wild.

I think in most use cases and agent would need just https and dns, both which can be MiTM monitored. In other some cases maybe also one or more of SSH, redis, MySQL, Postgres etc.

But YOLOing and letting it to connect to anything is probably not needed.

Maybe I missed it but when the booster was supposed to land or begin the landing burn it was lost by the cameras so it probably impacted somewhere other than where it was supposed to land.

It would be unlikely (not impossible) that board members would be briefed about ongoing criminal behaviour, and certainly not something so deep into operations as how the ECU is being programmed.

Can a board member be reasonably responsible for the actions of tens of thousands of employees if they have not explicitly enabled or condoned criminal behaviour?

The person that would benefit the most would be a senior executive who stands to gain a promotion, bonus or land an even better job elsewhere.

A former prime minister of my country was fined over $6 million for being on the board of a company what traded while insolvent. Not a prison sentence but a harsh penalty for someone that was not super rich (as far as I am aware).

> going all in on open source over the past 10-15 years.

Given that there are many Microsoft closed source extensions for VS Code, that cannot legally be used with the open source Version of VS Code, I would say they are not going all in. Knee deep maybe.

> High-strung, antisocial, egotistical, domineering, rage-filled.

I think you are doing the man a disservice summarising him in such a way.

His interest in unorthodox/heretical religion was at least since he was at university. He spent a significant amount of time on alchemy.

Newton was the President of the Royal Society for over two decades, an MP for a similar amount of time which I would think required a lot of interpersonal relationships and socialising.

He seemed to get along well with family who cared with and lived with him and described him as loving.

The traits of holding grudges and raging were probably as common in academia then as they are today (tech is benign in comparison), but are otherwise sociable and genuinely trying to be good, albeit flawed, people.

He made numerous statements of modesty, the most famous being "If I have seen further it is by standing on the shoulders of giants." This has, IMHO, been unfairly reinterpreted in recent times as being a insult to a rival rather than taken at face value.

If every comment, action, HN comment, tweet etc. of any person's entire life was interpreted in the least charitable light we would all be recorded in history as being as vile as you describe him.

I think at the end of the day he was just a gifted flawed human.

Thanks :) A/VR is not my thing. Hopefully they will do something in the future. Tertiary Education is in crisis thanks to AI, but it’s also going to supercharge it and that’s where research is needed.

> I do not know why concerts have to be SO LOUD. Loud, sure. Permanent ear damage loud, why

I have often wondered this. So it’s non-deafening loud at the back? I was at a concert recently that was way too loud. A sound guy came to check and stood in front of the speakers. I thought finally it’s going to be turned down … nope … clearly his hearing had already gone which would explain why it was so loud.

I think my hearing has been damaged in the past and so I now always have either AirPods at the very least or earplugs on hand. If anything loud, like heavy construction next to a bus stop, is happening I put them in. I can’t undo the past but I can prevent future damage.

> disclosure: I work at Meta as Llama Partner Engineer

Do you know of any programmes for Tertiary Education initiatives?

Jony Ive is a designer verses Tony Fadell who is a hardware guy.

Ive is a very talented artist but AI is not being held back by people unwilling to courageously make things thinner and thinner.

I would imagine Ive looked at an Apple HomePod and thought “we could make this beautifully flat and hang it on the wall of every room in the house”. This might be a good idea but it in no way solves the major problems with AI/LLMs.

I use it, I am required to use Windows, and it’s a huge improvement over doing Data Science on native Windows, but the terrible filesystem access ruins what otherwise would be a seamless experience.

It’s fine for running small models but when you get to large training sets that don’t fit in RAM it becomes miserable.

There is a line where the convenience of training or developing locally gives way to a larger on demand cloud VM, but on WSL the line is much closer.

It is disappointing that this requires rooting. Essentially this requires deciding if you want a dev environment or banking apps/nfc wallet or be willing to play an endless game of cat and mouse.

> Except with MCP, essentially most consumers will become "programmers"

You are more optimistic than I am. Most people I have seen are using LLMs are at best as an alternative to Grammarly or a document/web summary, or at worst making decision based on outdated LLM advice or as an inaccurate fact-engine.

The average person could code using Excel, but most don't even if they know how to use IF() and VLOOKUP().

> massive amounts of electricity and water ... we need to succeed on all fronts

Datacenters' used for training or batch jobs can be placed in places where water and power are plentiful, ambient temperature are relatively low and government/society are stable.

A datacenter is going to be built at the bottom of New Zealand where all these things are true. There must be plenty of places in the world where this holds true.

At least for training, I think it is possible to have our cake and eat it. For real-time inference, probably not.

I feel the points made in this article have been debated many times on HN.

The argument about power usage for developers, as opposed to consumers, is probably insignificant compared to the inefficient use of compute resources for deployed software today.

Arguably LLMs are probably enabling some web developers to create native applications rather than Electron monstrosities saving many P-core CPU cycles multiplied by the number of their users.

Optimising server applications with an LLM could eliminate unnecessary cloud servers.

Of course all the above could be done without LLMs, but LLMs can empower people to do this kind of work when they were not able before.

> Even adding some randomness to the timer tick just means you need more samples to suss the statistic out.

If each timer draws from the same random distribution then sure, you could work out the real tick with greater accuracy, but I don’t know if that is practical.

If the timers draw from different distributions then it is going to be much harder.

I imagine there is an upper limit of how much processing can be done per tick to before any attack becomes implausible.

Haven’t browsers closed that hole independently by reducing timing precision?

I need this! I am struggling to grow carnivorous plants in New Zealand. I think I am giving them what they supposedly need but I just can’t win.

I think I misinterpreted the map. So "low interference" on this map means "probably no interference" not "low levels of jamming"?

Because i agree with you, nobody is jamming on a large/noticeable scale in New Zealand.

I am quite skeptical at least with the low interference. My entire country, NZ, the whole of Europe, India, USA and most of inhabited Australia and Canada are reported having low Interference. Is this likely to be true?

In theory that sounds great, but most LLM providers are trying to produce useful models that ultimately will be widely used and make them money.

A model that is more correct but swears and insults the user won't sell. Likewise a model that gives criminal advice is likely to open the company up to lawsuits in certain countries.

A raw LLM might perform better on a benchmark but it will not sell well.

That is a really good question, I am not sure where to draw the line.

I think it would be safe to say BERT is/was firmly in the non-traditional side of NLP.

A variety of task specific RNN models preceded BERT, and RNN as a concept has been around for quite a long time, with the LSTM being more modern.

Maybe word2vec ushered in the end of traditional NLP and was simultaneously also the beginning of non-traditional NLP? Much like Newton has been said to be both the first scientist and also the last magician.

I find discussing these kind of questions with NLP academics to be awkward.

You read it correctly. I obviously didn't explain myself well.

I must have explained myself extremely poorly. I spent a fair bit of money ~$1,000 USD running a near SOTA fine-tuned llama model on cloud GPUs for this very particular task.

> using VADER over something like DistilBERT is NLP malpractice in 2025.

Ouch. Was that necessary?

I used $1000 worth of GPU credits and threw in VADER because it’s basically free both in time and credits.

I usually do this on large dataset out of pure interest in how it correlates with expensive methods on English language text.

I am well aware of how VADER works and its limitations, I am also aware of the limitations of all sentiment analysis.

I am in academia and worked in NLP although I would describe myself as NLP adjacent.

I can confirm LLMs have essentially confined a good chunk of historical research into the bin. I suspect there are probably still a few PhD students working on traditional methods knowing full well a layman can do better using the mobile ChatGPT app.

That said traditional NLP has its uses.

Using the VADER model for sentiment analysis while flawed is vastly cheaper than LLMs to get a general idea. Traditional NLP is suitable for many tasks people are now spending a lot of money asking GPT to do just because they know GPT.

I recently did an analysis on a large corpus and VADER was essentially free while the cloud costs to run a Llama based sentiment model was about $1000. I ran both because VADER costs nothing but minimal CPU time.

NLP can be wrong but it can’t be jailbroken and it won’t make stuff up.

Does the Zigbee work well/as expected? Does it have lower power draw when doing Zigbee vs. wifi?