(At the time of writing if you expand 'Data Products' the first file is 'Program for the International Student Assessment (PISA) 2018 Public Use File (PUF) NCES Number: 2021019 Release Date: July 8, 2021')
Rahim, you know this data better than me, but this purports to be it. Is it not real or is this somehow subtly different from what I would read it to be? It calls itself the reading results from PISA 2022 by race. You’d have to set a new dependent variable to get the other two. I don’t know how to get it to show all 3 at once.
I started programming at 18 right after highschool (which, I guess, is late by HN standards: a number of my peers had played with Python first and hated Ocaml).
He didn't say his IoT devices relied on 'cloud' accounts. He said there are IoT that use services on this domain.
Imagine he's set up some IoT devices at his parents' home, and those devices use services that he hosts somewhere on the internet. It would be silly to hard code the IP addresses in there, right (unless he operates his own ASN)? So he would use DNS to allow those devices to find his server(s). This would be the case whether the servers are at his home, at his office, or in a rack at a data centre.
I did something similar for structured text extraction. I added markers throughout each source document and then, for each piece of info I wanted, I asked the LLM to provide two separate fields:
xyz
xyz_citation
The latter was just the node number. So then my code could extract the exact snippet, instead of trusting the LLM to quote something verbatim.
I believe Gemini Flash is smart enough to know when to ground with web search. Their app has been saying it’s running a web search on almost all of my queries since 3.6. And given that Google … is Google, I trust them with web search grounding more than anyone else.
I wasn't trying to be precise originally, I just tried to fit activities into "morning / evening" buckets. I did the whole itinerary with Opus first, but when I gave it to Gemini 3.7 Flash to review, it started correcting it with "this place will close 5PM" or "this place is closed for good".
It was right on every nit, so it was surprising how well the model knows these things. If I ever release this I'll probably need the SERP API or Google Maps SDK (which I've heard is very expensive now), but for a personal trip where I will verify manually, using the LLM is okay for now.
When you called the Gemini API, did you opt in to using search grounding:
tools=[{"type": "google_search"}]
I'm curious whether in fact you were getting answers from the model weights (which is what I had assumed) or whether your API calls were resulting in web search tool calls.
Using grounding in Gemini is indeed backed by the same canonical data source for business information (like opening hours) as Google Maps. This stuff is available in its own API for a GCP fee, but we’ve built tooling to connect it to the Gemini agentic ecosystem as well.
So if I understand you correctly, Gemini has direct free access to the Google Maps API in a manner that others (people, LLMs) would need to sign up for API access and pay for?
Gemini models - at least via some interfaces - have tool calling API access to various Google integrations. flights.google.com, maps.google.com, etc.
The info isn't in the model weights.
Because of where I live, there are three viable airports for any given flight I might want to take, which historically has made shopping a real pain. But Gemini (and only Gemini) has greatly simplified it. Pramble plus date range plus destination and it very quickly generates potential itineraries with costs, total travel time (driving included), etc.
jampa was saying that the Gemini 3.7 (a model) ranks higher on real-world knowledge. The point I was originally making is that I would trust any model by itself to answer real-world knowledge problems. If I want to know opening hours, then probably any of the models could find the answer with a web search tool.
Perhaps the Gemini API makes this easier, as the web search tool is built in.
"It has led to academic improvements across entire school districts"
This sounds amazing! Is there anything more you can share about this? Or any articles/papers about the results or how you supported kids to progress with Scratch?
Nome, Alaska is a majority indigenous community as are many fly-in towns in Alaska, so you're basically saying something as tone-deaf as "why don't the indigenous people just abandon the place they've lived for thousands of years just because "modern" settled life showed up in the form of the United States of America".
Yeah, it's costly and difficult to live there, but that's their home.
These teenagers aren't thousands of years old. Neither are their parents.
The folks suggesting that other people should pay to improve things for people choosing to live in that area seem to simultaneously believe that these people:
- have a right to the benefits of what you call '"modern" settled life', AND
- cannot be inconvenienced by the things that make that life possible (like density)
> The folks suggesting that other people should pay to improve things for people choosing to live in that area
If the United States federal government wants to maintain a legitimate claim to sovereignty over much of rural Alaska, which is very sparsely populated, it needs to maintain an actual population there and provide basic services to it. It's a fairly well established geopolitical principle.
Similarly there's something like 41,000 people in Nunavut, and the Canadian government spends a shitload every year subsidizing all their communities' existence. But the alternative is worse, if you remove the population there's a danger that other interested parties might try to lay claim to it for themselves. It seems we're already seeing that with the US trying to claim Greenland for itself based on the concept that nobody is doing anything useful with it.
The ICJ said that sovereignty over "thinly populated or unsettled countries" can be sustained with very little actual exercise of sovereign authority. It upheld sovereignty over unsettled portions of Greenland. They later used the same principle talking about islands that were "uninhabited or not permanently inhabited".
You seem to be saying there's a general requirement that every part of sovereign territory must maintain a resident population AND that the government must provide certain 'basic services' there. But that doesn't seem to be true, at least according to the ICJ.
If what you said were true, then no state could claim sovereignty over an unhabited place. But that's not true. The ICJ ruled that Malaysia had sovereignty over Ligitan, even though the island has a population of zero. https://www.icj-cij.org/node/103163
I somehow doubt that the major world powers that might have the capability or inclination to occupy a depopulated Alaska or Nunavut would put much credence in or be scared by a strongly worded opinion issued by the ICJ. One of them is busy trying to invade their European neighbor right now with bloody results. The other one is on a major island building project in the South China Sea turning sandbars and coral reefs into air bases to claim the oceanic territory.
When you said "It's a fairly well established geopolitical principle." I assumed you were talking about the legal aspect. That's why I responded the way I did.
Now I understand you're talking about not about the legality, but the practicality. That's even easier to refute I think: if a world power wanted to annex that part of Alaska, the ~4000 current residents of Nome wouldn't be much of a deterrent, even with their hunting rifles.
No, they certainly wouldn't, but in the realpolitik of war, they would provide the justification for the USA to engage in an armed response to retain the territory. Basically think of the two scenarios of what would happen in the Falklands in 1982 if either:
a) the islands were claimed by the UK but also completely unpopulated and were occupied by Argentina
or
b) what actually happened, the islands were occupied by almost entirely UK-descent, English speaking subjects of the Crown, resulting in subsequent military action that had broad and overwhelming public support in the UK.
I was trying to find a comparison between Peekaboo (what openclaw uses for computer use) and cua (what Hermes uses). I didn't find anything useful on a quick google search, so I went to HN search and searched for 'peekaboo cua'.
I benchmarked Hunch against Peekaboo (OpenClaw's computer use) and Cua Drivers, and out of the 54 MacOS local tasks, Hunch completed 53 while costing $6.03. Peekaboo completed 41 tasks and used $31.03, and Cua Driver completed 31 tasks and used $37.42.
That makes me curious. But the repo only has 17 stars so I'm curious whether other folks on HN have tried it.
It sounds/looks good but I'd rather learn from someone else who has tried it, than run my own benchmarks.
Because it's obviously written by AI.
reply