I've been learning CAD, 3d printing, PCB design and brushing up on my embedded programming... all with the goal of being able to build toys for/with my son. It's incredible how accessible it is in todays world, made possible by these advancements:
- incredibly powerful and cheap microprocessors (esp-32)
- Fast, high precision desktop 3d printers
- Affordable small batch PCB manufacturing
- LLM's to advise on circuit design and help with embedded programming
Would you have any interest selling a non-comm license to the PCB, f3d files and source code? My 1.5yo son would absolutely love this!
Same, I recently got into Arduino so i could build a toy idea I had for my kid (a sort of "keepy-uppy" paddle game). I've always avoided Arduino projects because I don't like to code in my spare time (its my day job) and also learning how to wire things seemed daunting -- LLMs solved both of those problems for me (Claude's wiring tutorials are awesome and the code is simple enough that it can one-shot it).
I also bought a Bambu A1 3D Printer and it is unexpectedly way more fun and useful than I thought it would be. I designed the toy in TinkerCAD and it printed out beautifully (I also have been printing out lots of other toys and yes, useful things for around the house and for other projects).
Next steps are learning Fusion 360 and figuring out PCBs -- That also seemed daunting to me but its nice to see other amateur hobbyists are seemingly picking it up with not much difficulty.
I highly recommend the "learn fusion 360 in 20 days" series on YouTube. I think I did it pretty causally in half that time. Then just find excuses to design stuff. After about a year of random projects, I can pretty proficiency draw most simple assemblies. This morning my son broke the battery hatch on his favorite toy and within an hour I had a replacement printed. (ducktape would have certainly been faster but where's the fun in that ;)
One thing I would point out is if your interested in organic shapes that blender is the better tool to learn/use. Fusions surface modeling is pretty hard to work with.
This response is spot on. People seem very confused about what MCP actually is. It's just a standard way to provide an LLM with tools. And even how that happens is up to the agent implementation. There are some other less common features, but the core is just about providing tool definitions and handling the tool_call. Useful but basically just OpenAPI for LLM
I think people are really underappreciating the "OpenAPI for LLM" part. The hype forced a lot of different SaaS products and vendors of all stripes to actually follow a standard and think somewhat critically about the usability of what they expose.
At least 50% of the YouTube promoted videos I get are crypto currency scams where some paid actor walks you though deploying an eth contract that empties your wallet. I report every one and nothing changes :(
I get 50% AI generated tai chi promising strength gain, weight loss and enlightenment, the other 50% israel sponsored ads assuring me people in gaza are not starving at all and completely healthy
There were many other similar ones, especially on the smaller digital ad spaces that were basically just TVs on the side of roads. And those ones were more specifically calling for empathy for the deaths occurring to Israeli people during the war on Palestine
I pay for YouTube so all I get is paid creator promotions for VPNs and Squarespace unless it's someone being sent a free thing in exchange for a review
Normalize paying for things instead of selling your attention to the highest bidder.
Normalize paying for things instead of selling your attention to the highest bidder.
But you just admitted that you pay for YouTube and it shows you creator promotions. You are literally paying to see ads, then telling people not to do the same.
Unless there's some subtlety I'm missing here. I haven't been on YouTube in at least a decade. I see no difference between a blogger pushing a VPN and Google showing an ad for a VPN.
The big draw for cable TV was that you could watch TV without ads. Then ads started appearing on cable and people said it's OK, because the content is higher quality and not available elsewhere. Then that changed, and now there is no difference between broadcast, cable/satellite, and streaming services. Except that you don't have to pay for broadcast. (Yet. It's coming.)
> Unless there's some subtlety I'm missing here. I haven't been on YouTube in at least a decade.
Youtube Premium is fighting back against the sponsor segments with this "commonly skipped segment" feature. You hit a fast forward button and it automatically skips ahead to the place most people jumped to.
A year or two ago somebody asked Adam Ragusea about whether this type of skipping causes problems for creators - and what he said was basically that if viewers see the brand name / call to action at the end of the ad, that's mostly what matters to sponsors.
No idea if that's been borne out in practice, though.
If you don't like youtubers with sponsors, don't watch those videos. Not all do.
Personally I pay for youtube and I don't mind the sponsor sections. They're easy to fast forward through and income goes directly to the creator. Youtube doesn't take a cut. These are the only kinds of ads that work on me - in the rare case that the product is something I'm interested in, I go out of my way to make sure I use the creator's link.
The long story short is that there are creators I like and I want them to devote all their time to making more content. I'm glad some of them get sponsors. For many I just straight up give them money on Patreon.
I've gotten rid of 90% of the ads by paying for YouTube, the rest of the ads I skip by jumping forward in the video which is annoying but only a little OR by being legitimately interested in what the person has to say if they're reviewing a product which has been in some way paid for. I'm also just fine with someone promoting their own merch or patreon which I am sometimes actually interested in.
The subtlety I don't get why you're missing is I now have very much reduced ad exposure and the rest I do have is entirely controllable.
I choose to not watch YouTube. I was born in the 80s, am a software engineer, and I’ve watched maybe 10 total hours of YouTube, all 100% of those hours were car-repair related.
None of those scams were intermixed with popular legitimate content. If Facebook had a tab at the top called "Scams and Other Nonsense" and you clicked on that and it had a bunch of scammy content, that would be an equivalent to late night infomercials. But Meta doesn't do that. It mixes the scams in with all of its popular and non-scam content so you cannot easily tell it's scammy. Worse, it targets people vulnerable to those scams by tapping directly into their interests and sentiments in a way TV never could.
You are making a silly argument here. There's no equivalence at all.
Mostly, I'm getting things like German ads for my local German supermarket (that I would've gone to anyway without the ad) dubbed badly into English with an AI that can't tell how to pronounce the "." in a price, plus a Berlin-specific "pay less rent" company that I couldn't use even if I wanted to because I don't rent.
But when I get 30 seconds of ads a minute into a video that had 30 seconds of ads before I could start watching… I don't care what the rest of the video was going to be about, I don't want to waste my life with a 30:60:30:… pattern of adverts and "content" whose sole real purpose is now to keep me engaged with the adverts. (This is also half of why I don't bother going to Facebook, every third post is an ad, although those ads can't even tell if I'm a boy or a girl, which language I speak, nor what my nationality is, and the first-party suggested groups are just as bad but grosser as they recently suggested I join groups for granny dating, zit popping, and Elon Musk).
> I never understand why well-paid HN commentators refuse to pay for their entertainment.
People don't want to pay (help) people they don't like. YouTube ads do not feel fair, they feel manipulative and unethical. It's expected that most people wouldn't want to willingly engage with that kind of asshattery.
Contrast that with platforms like twitch. I'd say the average twitch viewer (that interacts with streams/chats) has a slightly negative view of Twitch. But many will still willingly donate dozens of subs to streamers they like. This removes ads for other people, not themselves.
People think YouTube is greedy and untrustworthy. Why would you willingly feed that machine?
I don't pay because it feels like paying protection money to the mafia. "Here's an annoyance/danger we created for you. If you pay us, we'll stop doing it."
on the other hand, it's how their business model is able to work? People get wayyyy more views on YouTube than they do on Patreon or federated platforms or Nebula or Floatplane or or or or or or or
You don't see any adverts on youtube if you don't watch youtube.
My local cinema charges me to watch a film. Sure I could sneak in through the fire escape.
That's not paying protection money.
If they say "you can watch for free if you attend our timeshare presentation first", then that's still not paying protection money.
If I want a magician to entertain me at a party, it's not paying protection money. If they say "you can watch for free but only if you listen to me drivel on about some cause first" that doesn't either.
People on HN are unwiling to pay people to entertain them. Its astounding.
I think it really depends on how much you use it. For example, there is no way that I would pay for Facebook. It annoys me greatly that I’m forced to use it a few times per year, and I have to sell all of my data for it, but unfortunately I don’t pay just to avoid data gathering about me, because it happens anyway, no matter what I do.
But I pay happily for YouTube, because I use it daily, and my home country’s propaganda was annoying enough to make it worth.
The recent YouTube updates made me disabled my adblocker for a bit. Almost every ad I got was a scam. I reported one to YouTube, a deepfake investment scam. They actually got back to me and said they had removed the ad.
YouTube Premium Lite used to exist years ago, then they discontinued it in 2024 (I know because I used to be a very happy subscriber), now they brought it back but only in a few selected markets[0].
Google products' bullshit as usual, I never needed/wanted YouTube Music and the other bloat they wanted to force me to pay for, I was happily paying to not have ads...
Changed to what? Should dang become legally responsible for any of the bad legal advice I've been giving people on this forum? Should Murdoch go to prison for the lies in the paid advertising that Fox anchors and opinion wonks are doing every day?
Let me take things back a step - it's nearly impossible to hold people who are lying accountable. Surely the platform bears less responsibility than the liars on it?
I don't know but I think there is a room for compromise. If you post illegal things online and the site cannot identify you so that you can be held accountable then the site should be held accountable. As it stands people are harmed and nobody is liable so we end up in this situation.
There's a simple way to do that. Legislate a requirement for ID, the users will provide it to the platform, the platform will provide it to law enforcement when requested.
Kind of like how South Korea (where you need a national ID to access digital services) is doing, or the UK is trying to do with their ID push.
(And then who wants this could go have a fight with the people who don't want this.)
Speeding in a rental car doesn't absolve you of the ticket.
But I mean yeah, you could theoretically rent a car in the UK and kill somebody then flee to the US and hide. But don't let perfect be the enemy of good.
HN users used to herald that law as the best thing since Betty White (who was older than sliced bread).
Without 230, there would be no YouTube. No Facebook. No Instagram. No social media. Forums would likely have gone extinct. Half of the tech industry and a good chunk of the jobs that people on HN do would never have happened.
Now people on HN want to get rid of the law. People who are too young to know what it was like before that protection set the internet free to create and collaborate.
I despise social media. But demonizing 230 just shows a basic lack of knowledge of history, economics, and the reasons it was created.
Well, give us the argument, then, instead of the mere allegation that history is frowning at us. Why is it not possible to change the law to permit platforms to not be liable for speech of their users, particularly when users are engaging in a platform in the capacity of communicating and exchanging information, (i.e., 230 as it is today) but not permit advertisers from displaying ads which contain blatant fraud, for which the advertising platform is profiting off that fraud?
Usenet was a thing. A huge thing. There was zero danger that it would go "extinct" due to the lack of extralegal protections.
> But demonizing 230 just shows a basic lack of knowledge of history, economics, and the reasons it was created.
You're ignoring the context. Section 230 was created when the Internet was nascent and we were trying to encourage broader /business/ investment into the technology.
Now that that investment has occurred and most consumers _prefer_ to do business on the Internet, whereas the opposite was previously true, we no longer need the _additional_ protections for hugely profitable businesses.
Aside from that is there some reason we can't _modify_ the law to bring it more in line with citizen expectations? We're bound to the decisions of the past absolutely? Please...
I dunno,I saw a video of mister Elon Musk himself telling me without twitching a muscle in his face except his lips to put all my money on his new crypto venture. Seems pretty legit to me.
Yeah, it's wild how poorly the hackernewses understand this. If the ad platform has few signals for targeting, but it does have the available signals of you're using a weird VPN or tor, and a weird user agent on an uncommon platform, then it's just going to assume you're a crypto loser like the other people sharing those traits.
For what it's worth, I see no crypto videos. YouTube recommends stuff I find enjoyable (lots of sketch comedy, TTRPG videos, interesting documentary style stuff, BTS on video game development, etc). I really have to wonder if your tastes align with crypto currency scams.
That being said, I am paying for Premium, so I wonder if you are, and if you are blocking ads.
We are talking about ads and promoted videos. Nothing to do with what it is recommending unless I am entirely conflating the root of this subject. If that is true, then of course you would never have seen these as a premium user.
Scam videos are the chum box ads of the video world. Usually the lowest cost ads and so if you block tracking or are viewing a video in a private session you will have the highest chance of hitting these ads.
And? YouTube web absolutely has ads and if they have not built a model on your user you will absolutely get the chum ads like scams. I am not sure what you’re trying to tell us.
I wonder if it matters that I'm not signed in on my Chromecast youtube app, but am on the app I'm casting from. That I don't get ads for nerd stuff seems to imply it's not using data from my account
Gotcha. So you are ignorant of why people are commenting.
The OP was talking about seeing 50% scam crypto ads. Our responses were to provide a comparison. Not to say that it doesn't happen, but that 50% scam crypto ads are not the norm for everyone. It's helpful to have that comparison when providing anecdotal information.
No one is saying those ads don't happen, only that it's probably not normal.
Next time, instead of being unnecessarily antagonistic, admit to being ignorant and ask.
Please don’t start drama where nothing exists. You were confused and I pointed out that 1) I believe we are talking about promoted, which is paid, videos not the recommendation engine. These of course are not purely ads but are paid for. And 2) that these chum style ads and promoted videos have a much higher prevalence with folks that block tracker where user profiles have not been built. It’s the chum ads of the video world.
I am simply asking what is the point of the response to my comment. Ads of all degree exist but these scams do exist in a pretty large % of the ads shown but perhaps much lower dollar value since they get shown to profiles without a tangible viewer model.
Next time, instead of using inflammatory language please just slow down and reread or have a more thoughtful discussion. Thanks.
You’re arguing over nothing. All I have consistently said is scam ads/promoted videos exist in high volume on YouTube but usually only for viewer profiles where YouTube has little to no profile. Those are the cheaper spend areas hence scam ads. YouTube like most of the other platforms do little to police it.
The ads are get rich quick, unregulated powders, crypto etc.
So again I am not even sure what you are arguing about or why you use such nasty language but read what I keep saying and relax.
Ad and "promoted" videos are different in this context. And the OP was mentioning promoted videos, not ads.
> At least 50% of the YouTube promoted videos
I've never seen a "promoted video" (whatever that is specfically) that deals with crypto. Note: Premium users can still see promoted videos. I imagine these are more targetted to people who would want to watch these sorts of videos.
> Nothing to do with what it is recommending unless I am entirely conflating the root of this subject.
I was referring to recommended not in a strictly technical sense, but in a way any normal person would use the term. e.g. Recommended videos meaning: All the videos youtube shows me that it thinks I might want to watch. Whether these are officially "Recommended" or "Subscribed" or "Promoted" or whatever, I don't know.
What I do know is that I don't see any crypto scam videos or ads.
> If that is true, then of course you would never have seen these as a premium user.
Apparently, that's not the case.
tl;dr: We are talking about videos like normal people. You are wrong.
No... these are "paid promoted" videos that show up in your feed[0]. They are different from ads that roll when a video is playing. Example screenshot I found on reddit [1].
- video from screenshot[2]
- coe from video[3]
I'm guessing I get served these because I typically interact with them because I'm curious to read the code they link to see how obvious the scam is. It's also fun to reverse face search the actors and find them on fiverr.
As I originally stated I am happy to be corrected but I don’t think you understand it either. Promoted videos are ads. Premium removes in-stream ads, not in-feed ones. They’re paid placements, not recommendations.
And again as a premium user you won’t see chum style feed or promoted videos because premium removed the feed style and promoted will be more tailored to your preferences.
Which coming full circle leads us back to my original statement. If they don’t have a good user profile for you, you will get lower cost ads (promoted videos) which generally are going to be the chum box of ads, crypto, magic formula powders, get rich quick.
"Code interpreters" are incredibly powerful tools for agents as it allows them to process large amounts of data without actually having to move the tokens through it's context window.
However, I don't actually see what any of this has to do with MCP. It's more-so just tool calling + code interpreter design patterns. If anything, the MCP hype has resulted in a lot of BAD tools being written that return ridiculous number of tokens.
SQL is really a perfect solution for allowing the agent to access data, but in most applications, it's not realistic to provide it with a db connection. Your either need RLS and user connection pooling (like supabase) or strict application tenant filtering (which is tricky) and even then, you still can't efficiently join data from multiple sources.
I recently built a system using tenant isolated S3 "dataponds", populated with parquet files, that the agent queries with duckdb.
The VM (agent core) gets an short lived "assume role" STS token injected so it can only access buckets the user is entitled to (also VPC networking so no other ingress/egress). Each location has a `manifest.json` file that describes the contents and schema. The agent can query the parquet files using duckdb, do additional processing with pandas and then share files back to user by writing it to S3 in special predefined display formats (data-table, time-series, etc).
The file ids, along with a sample of the data is pass back the LLM. It can then embed the file in it's response using a special tag and a custom markdown renderer displays it back to the user.
From what I can tell, this is basically what Chat-GPT has done for a long time (minus the preconfigured "datapond" and special response formats), but it's pretty awesome how "low-effort" a system like this can be built today.
100% same experience. If it were up to me, I'd started with typescript, but the client insisted on using a python stack (landed on FastMCP, FastAPI, PydanticAI).
While, `PydanticAI` does the best it can with a limited type system, it just can't match the productivity of typescript.
And I still can't believe what a mess async python is. The worst thing we've encountered was a bug from mixing anyio with asyncio which resulted in our ECS container getting it's CPU pinned to 100% [1]. And constantly running into issue with libraries not handling task cancellation properly.
I get that python has captured the ML ecosystem, but these agent systems are just API calls and parsing json...
async python has problems, but "anyio exists" is not one of them that can be blamed on python, simply dont use weird third party libraries trying to second guess the asyncio architecture
edit: ironically I'm the author of a weird third party library trying to second guess the asyncio architecture but mine is goodhttps://awaitlet.sqlalchemy.org/en/latest/ (but I'll likely be retiring it in the coming year due to lack of interest)
I don't recall the exact situation but am I suppose to just know which async library each dependency is using? It reminds of of the early days of promises in JavaScript.
This is how MCP works if you use it for as essential an internal tool API gateway (stateless http) instead of a client facing service that end users are connecting directly to. It's basically just OpenAPI but slightly more tuned for LLM inference.
Nvidia is basically allowing some of its clients to pay with partial ownership of their companies instead of with cash for the chips they want, because there's simply not enough cash in them for the lofty goals lots of people think are achievable.
God that sounds so crazy. Its giving me "sell our municipal water to Private Equity to buy a park, imagine the tax revenue!" vibes except its just diluting the shares.
It feels like economics in the US is turning into FanDuel mobile gambling. Can we pass a law to disincentivize this garbage and make them start paying dividends again?
It's more like "I'm making so much bread and have so much money already. You are making clothes but don't have enough money to buy bread from me to feed your workers so you can't grow. I think our village population will soon explode and there's going to be huge need for your clothing. So how about I give you money to buy bread from me in exchange for partial ownership of your business and its future profits?" It's a perfectly reasonable thing a rich miller could do for a poor but growing village businesses.
No ... its more like "I'm making so many GPUs and have so much money already. You don't have GPUs so how about I give you GPUs in exchange for partial ownership in your business so you don't have to put this capex on your books that affect your ability to raise other investments, trigger certain disclosures, or make your finances look underwater since the assets will likely bolster your own valuation."
Feels gamble-y when the shovel maker is playing weird VC for the gold prospectors like this, but I get the appeal since Nvidia stands to lose everything if this stuff tanks anyhow and it further entrenches Nvidia's own position all while keeping the hype train steaming by moving units.
That is an accurate description. It's a gamble that Nvidia takes. Just making chips would be a normal bet. Making chips and getting paid with equity in its customers is like a leveraged bet. If the AI succeeds Nvidia will own significant vertical portion of all of it. If it fails, Nvidia will shrivel. But by deciding to take that risk Nividia tips the scale on its bet making AI more likely to succeed (or just not to fail because of chip and capital starvation).
Do you think its smart to structure commercial ventures this way? In a different space, is it a great idea for a farmer to sell %10 of their venture to John Deere for the tractor?
If otherwise farmer would not be able to afford the tractor and starve instead I'd say it's pretty reasonable.
You can fund your education with part of your future job earnings.
In a perfect world companies should just do their business and nothing else. Companies shouldn't be able to own other companies because that's gambling on their value. They shouldn't be allowed to do stock buybacks because that's gambling on their own value (and a kind of insider trading).
In our world however companies are free to gamble on the side as long as their owners don't mind.
Whether what Nvidia does is the best or the worst idea, only time will tell.
On the plus side, you can just declare bankruptcy and fold instead of becoming a share cropper.
Heres the thing, these assets deprecate, while the stock doesn't. I see a big asymmetry in deals like this, and I do not like it. When you start a business and become indispensable to your customers, you're doing well. If a company fails, it is maybe unavoidable. In situations like this, the risk increases because now these companies are interdependent not only as suppliers and providers, but in their intrinsic value and likely decision making insofar as shares can be voting shares. In any case, the C Suite and board sure know who owns %10 of the shares.
>In our world however companies are free to gamble on the side as long as their owners don't mind.
This constitutes an enormous systemic risk. We are treading in dangerous waters. The owners are not the only stakeholders in a company.
>In our world however companies are free to gamble on the side as long as their owners don't mind.
> This constitutes an enormous systemic risk.
Tell me about it. 2008 crash was a result of exactly that. Banks instead of doing banking focused on gambling on risky assets and lost. Nvidia gamble is looking less obviously bad.
Nvidia can do that because what they sell has an insane markup. Nvidia is getting a billion dollars in equity for input spending of a hundred million. Nvidia is creating demand for it’s products and in turn getting ownership in companies for cheap. It’s like being a VC but your investments are chips and the chips also count as revenue for you.
At least investing in companies that will buy Nvidia makes sense somewhat, even if it feels dodgy. What possible sense does it make for Nvidia to invest in a maker of telecom equipment (Nokia) with no public plan, and the value of the investment jumps 30% as soon as it’s announced? It’s not like call routing needs to be “AI enabled” or they’re putting Nvidia GPUs in their custom silicon.
Feels like Nvidia has too much money and they know whatever they invest in will appreciate thanks to the “AI premium” that investors bestow.
Nokia had some of the best mobile hardware (and software!) in the game until they were bought by Microsoft and promptly killed.
I've been patiently waiting for their resurgence. Building embedded/mobile devices is their forte, which is also coincidentally the hardware space where AI is most poised to shine. Nvidia made a move in the same direction when they tried to acquire ARM. That got antitrusted (fairly so) but investing in a now-decimated telecom company isn't likely to ruffle any feathers this way.
> I've been patiently waiting for their resurgence. Building embedded/mobile devices is their forte
Wouldn't most of those people have gone elsewhere by now? If you're a mobile device superstar, why would you stick around at Nokia once the mobile device part of it crashed and burned?
A company's just a legal structure, people change over time. And that was more than 10 years ago, and didn't they sell their mobile division to Microsoft?
(Nokia went from king to pauper in the year when the iPhone and then Android launched. They death spiralled and Elop was brought in a few years later to transition them to using Microsoft Windows mobile. However, windows mobile was as old and uncompetitive as Symbian so couldn’t compete. Was very obvious at the time although the boards seem to have been living in their dream worlds)
There is a public plan (literal press releases) and it is to use Nvidia Grace chips in Nokia’s future RAN products. Seems to have cooling issues but seems to be a nice chip for the most part.
Nvidia quietly in the background is getting more interested in the network side of things and are hiring accordingly. This is an area I don’t really think you have all the details to have an opinion on imo.
At 18, I had no business even being responsible for a kid for a day. I hope the stability (both mentally and financially) that I'm able to offer my kid in their development years far outweighs the fact I won't be around as long.
Might this not be a consequence of how "we" were raised? In the past there were 18 year olds who were successful and responsible leaders of nations. Marcus Aurelius became emperor at 16! My parent just wanted me to go to college and earn money so I grew up with a rather 1-dimensional and hedonistic perspective on life.
I'll be encouraging my children to have children as early as reasonably possible. In part it's because of greed - I want to see those grandkids, and maybe even remotely possibly great grandkids before I die, but it's also because it's what I wish I had done in hindsight. Having children has not only been the joy and pride of my life, but it also gave my life much more meaning and direction. In any case, I also think it's completely appropriate for grandparents to play a significant role in the raising of children.
- incredibly powerful and cheap microprocessors (esp-32) - Fast, high precision desktop 3d printers - Affordable small batch PCB manufacturing - LLM's to advise on circuit design and help with embedded programming
Would you have any interest selling a non-comm license to the PCB, f3d files and source code? My 1.5yo son would absolutely love this!
reply