Hacker Newsnew | past | comments | ask | show | jobs | submit | chhxdjsj's commentslogin

I wish someone would do a large RCT of water fluoridation in pregnant women looking at long term cognitive outcomes if fetuses. It would be an easy study to do (just randomize each group to receive free deliveries of either fluoridated or not fluoridated water) and then look at their offspring’s scores on cognitive tests every few years. I think reputable scientists don’t touch this because they’re afraid of being labelled kooky.

Does the word Tuskegee mean anything to you? Because that's basically what you are asking for.

Nor is there any need for such a study as we have a natural one: some areas have more fluorine in the water than others. We started putting fluorine in the water because we noticed that the places with higher natural levels had better teeth. There comes a point where it's too much and downsides appear, but, again, we already knew that. Note that this is a completely normal thing--there is nothing which is not toxic in sufficient quantity. Including *everything* that is necessary for human life. What would be strange is if there wasn't some maximum safe level. Some things the body easily eliminates and the range between minimum and maximum is quite wide. Things which are not so easily eliminated have narrower ranges. Thus we have the situation where overdose of water-soluble vitamins is basically unheard-of, but overdose of fat-soluble vitamins very definitely happens.


I just do everything i can around the house immediately (as long as I dont have to go to work or whatever).. if its a long job it gives me the opportunity to listen to a podcast. “industry is its own reward”.


+1 to using it as opportunity to listen to a podcast.

I consider it some of my best podcast listening time. Somehow I’m more tuned in when doing house chores vs. trying to just sit and listen, or while walking outside where I have to pay attention to traffic when crossing the street and such.


Anyone got any tips on ways to buy ipos/equity on Shanghai STAR exchange? I want unitree and cxmt


You cannot buy these. You can only buy companies listed in Shanghai and Shenzhen that are enrolled in Stock Connect, which has certain requirements that means these small IPOd companies cannot be enrolled. It is possible that there mutual funds or ETFs in HK that are available, and hold these types of companies.


Me too - these are incredibly interesting companies and the valuations are modest.

Unfortunately as far as I can see it is not possible from where I am (Denmark).

Some Hong Kong shares and linked Shanghai shares are available in interactive brokers but not these ipoes.


"actually, did we say we had 8,000 tons of gold in reserve? we meant 9,000. Yeah we just counted, no you can't look at it, we have 9,000, here take the yuan and go build vaccines"

You cant do that with bitcoin


Sure you can. "No you can't look" is the key. All those blocks which hasn't moved in many years? They're all ours. Pinky promise.


Imagine if there was a gold which could be transferred across borders easily and completely securely within seconds, proof of holdings and ownership could be easily proven, and there was no future risk of gold supply shocks (eg gold asteroid hitting earth or alchemy becomes viable).


This sounds amazing! Is it also protected from being lost forever by trivial mistakes that are very common?


Yes, when managed by someone competent, just like Gold.


Has anyone had experienced with the giant magnet tiles? This article seems like suspicious affiliate marketing which places are beloved product that everybody has beside one that looks kind of similar.


Author here, I don't think I did affiliate market links.


First I’m seeing them, but that means very little. Every preschool we used had the normal magnet tiles though so I’m familiar with the concept.

Fair to question this. Most things are fake and… lame.


just encountered these in a local public library a week ago. toddlers loved them.


An excellent use case for this is ethereum smart contract verification. People store millions of dollars in smart contracts that are probably a one or two iterations of claude or gemini away from being pwned.


How did I know this was an israeli company just by how unethical they are at scale?


Well, you’d be surprised to discover that Koi is also an Israeli company, and they were the ones who even discovered this

https://www.calcalistech.com/ctechnews/article/syoe1xjslx


It would have been no less suprising to me had it been a US company but it certainly fits the cultural stereotype of callousness that particular country has been openly displaying in recent years.


And what are the odds that mossad are getting access to this data?


Not so relevant to the thread but ive been uploading screenshots from citrix guis and asking qwen3-vl for the appropriate next action eg Mouseclick, and while it knows what to click it struggles to accurately return which pixel coordinates to click. Anyone know a way to get accurate pixel coordinates returned?


How do you prompt the model? In my experience, Qwen3-VL models have very accurate grounding capabilities (I’ve tested Qwen3-VL-30B-A3B-Instruct, Qwen3-VL-30B-A3B-Thinking, and Qwen3-VL-235B-A22B-Thinking-FP8).

Note that the returned values are not direct pixel coordinates. Instead, they are normalized to a 0–1000 range. For example, if you ask for a bounding box, the model might output:

```json [ {"bbox_2d": [217, 112, 920, 956], "label": "cat"} ] ```

Here, the values represent [x_min, y_min, x_max, y_max]. To convert these to pixel coordinates, use:

[x_min / 1000 * image_width, y_min / 1000 * image_height, x_max / 1000 * image_width, y_max / 1000 * image_height]

Also, if you’re running the model with vLLM > 0.11.0, you might be hitting this bug: https://github.com/vllm-project/vllm/issues/29595


Will give this a go, cheers :)


It’s been about a year since I looked into this sort of thing, but molmo will give you x,y coordinates. I hacked together a project about it. I also think Microsoft’s omniparser is good at finding coordinates too.

https://huggingface.co/allenai/Molmo-7B-D-0924

https://github.com/logankeenan/george

https://github.com/microsoft/OmniParser


Thanks ill try this!


Could you combine it with a classic OCR segmentation process, so that along with the image you also provide box coordinates of each string?


It's very not accurate, but sometimes instructing to return pyautogui code works.

  prompt: I attach a screenshot (1920x1080). Write code to click the submit button using pyautogui.
  attachment: <screenshot>
  reply:
    import pyautogui
    pyautogui.click(100, 200)


Ive been asking for pyautogui output already but it is still very hit and miss


Also curious about this. I tried https://moondream.ai/ as well for this task and it felt still far from being bulletproof.


you want get the exact coordinated by running a key point network to pinpoint which coordinates does the next click point is you can. here I show a example simple prompt which returns the keypoint location of the next botton to click and visually localize the point with a keypoint in the image

https://chat.vlm.run/c/e12f0153-7121-4599-9eb9-cd8c60bbbd69


HFSP


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: