emacdona

Same! I have a huge soft spot for CF. It put a lot of food on my table in the early 2000’s. Then I switched to Java :-)

> I'm curious on why you think that, say, an M2 running linux will ever be a better choice for AI workloads than an M5 Pro running MacOS?

I mean that if Asahi starts to gain on Apple with respect to generation of M chip they support. If Asahi supported M4 chips...

But clearly, my dream hasn't come true yet :-)

> What is it that you think is in the hardware that you'd be able to unlock with linux but isn't exposed by MacOS?

I believe that Linux would give me more control over exactly what is running on the machine (and using up precioius memory). Maybe it would also have better support for the toolchain/ecosystem needed to run local inference?

I have this dream that previous gen M chips will be worth more than current gen M chips b/c people want to run local inference on Apple silicon... but they want to use Linux and not OS X.

That's the optimistic take. I have a much more pessimistic take: there is now only one place where people who want to sell dental devices have to focus their efforts. Less effort for them... and because of the scale... much bigger payoff.

When I was thinking of the examples I gave (poison the training and/or pay the providers), I was intentionally drawing analogs to how I saw Google search decline.

First, SEO "experts" poisoned the results. Then Google saw how much money could be made, and they sold the ad space directly to marketers. If you search for "Best <widget> for <doing something>" on Google, I am certain you are not getting the "correct" answer. And Google has suffered no consequences for this. To the contrary: it has been extremely lucrative.

> In the future, you will go to a special dental imaging facility where a modestly-compensated technician images your whole mouth and feeds the results into an AI system, and when you get home you can look at the assessment and chat with the AI about it. Then you can potentially go visit a dentist, oral surgeon, etc, if you actually need work. This system will produce far better dental health outcomes at far lower cost: the diagnoses will be more accurate, and there will be no bias towards additional, unnecessary procedures.

Right up until the sellers of dental supplies find a way to...

Indirectly influence model training to increase the frequency with which models recommend dental work and/or lower the bar for what they determine to be "necessary"...

or...

Directly influence model recommendations by paying model providers (who will be happy to take that money).

The point being that models (and particularly the humans that control them) are presumably just as corruptible as the humans they are replacing.

Wow, the timing of this question is eerie. Just last week, I got tired of talking about segmenting my network but never doing it. So I did some research, asked some friends, and settled on the following:

* An Opnsense firewall (purchased from Deciso; ie: "official" hardware)

* An OpenWRT One (ie: "official" hardware) to be used as an access point

* A Microtik switch (SW OS, not Router OS)

My plan is to create multiple SSIDs, have each one on a different vlan, and then control services provided to that vlan. For example, I'll dump all IOT crap onto a vlan that doesn't have internet access.

I'm still not 100% sure it will work (the hardware still hasn't arrived). But finally, to answer your question: "Yes!"

Holy crap, give me official hardware. I installed DD-WRT on my linsys router. It's been so long that I don't remember exact details, but the process was INTIMIDATING. I felt like I had to download a blob from a website that wasn't the official website (ie: I seem to recall different people building DD-WRT for different hardware). I also recall some features not being available on some hardware.

Give me the hardware that is sold by the people that work on the project!

I want some reassurance that I have the "most supported hardware", so I don't have to go through huge feature matrices to see if my hardware supports feature X.

I want headroom to install new packages. I don't want to have so little space that I need to remove one package to install another.

And finally, I want the people writing the software to GET PAID, because I want it to still be around... and being actively developed 10 years from now.

PS:

I initially wanted to get a consumer grade switch, but my network operator friends told me that was a terrible idea... and that I should get a use Cisco switch on Ebay. Cisco is probably great and network operators love it... but holy crap... trying to figure out what switch to buy is an enormous pain in the a$$. I pushed my friends, they relented and said: "Mikrotik is also great". I went to their website. 1000 times better than Cisco's; their switches were cheaper. I didn't have to "call for a quote" or scour Ebay. I haven't turned it on yet, but the buying experience was so much better.

I lol’d when they said “there’s no time to waste”… after making me watch five splash screens.

Thank you for making this. Biting commentary on modern video games.

Check your public library. If you are lucky enough to live near a big one, they may have all the shows you want to watch on DVD already.

I live in a large metropolitan area and belong to 4 different public library systems. Two of them have branches within walking distance of my house. For either of them, I can go to their website, search for a dvd, and have it delivered (for free) to the branch closest to my house.

It's not instant gratification -- it takes a couple of days. But in some ways, that's a feature, not a bug.

On top of that, they also have Kanopy (a streaming service offered via libraries). The selection is NO WHERE NEAR as complete as major streaming services... but, it's certainly nice to have.

Thank goodness... I thought I was the only one complaining. Ever since Apple started putting ads in the App store, I started referring to them as an "ad company". Every time I search for, say, "Signal" and see the first result be a "promoted" link to "Whatsapp", my blood boils.

F*k any company that shows me ads on the hardware/software that I bought from them. ESPECIALLY if they were once NOT an ad company... and then transitioned to one.

I'm fine seeing ads if they were part of the original deal -- for example, the ad subsidized Amazon Kindles. But nothing is safe or sacred anymore. I am all but certain that if you paid for a Kindle with no-ads (ie: you paid the premium), you can be 100% sure that you will eventually see ads!

All contracts, all privacy policies, all terms of use... come with a clause to the effect of "we can change this whenever we want and by using it, you consent!"

Mark my word: very soon, every commercial OS will use shady TOS to train their AI models on every byte you have on your hard drive. ZERO privacy. And if Linux becomes popular enough, you can be certain that the ad companies will start financing open source projects so that they can pressure them to exfiltrate data.

I remember when I was young, there was an organization, Nielsen Media Research [1], that would seek a statistically significant number of volunteers whose television viewing habits they could monitor. They existed because it was ILLEGAL for cable companies to monitor and collect data about what you watched.

Others have said it, and I'll repeat it: this is NOT a problem that can be fixed with technology. People need to get pissed off and write their legislators, and those legislators need to grow a backbone.

In the meantime, the ad industry lobbyists will be spending tons of money to (1) convince legislators otherwise and (2) keep the masses ignorant of the scale of the surveillance apparatus they have created.

[1] https://en.wikipedia.org/wiki/Nielsen_Media_Research

I used to be one of those people that would countdown the hours to Apple's launch events and tune in for the live stream. I was never someone who updated my iPhone every year, but I _did_ update my iPhone every two years.

Now, I can't even remember what version iPhone I have. I just remember that I bought it after my last one stopped working (after maybe 5 or 6 years of having it). I won't lie: part of the reason is that I'm just getting older and care less about these things.

But a bigger part is that new tech just isn't fun anymore...

I _hate_ using my current iPhone for surfing the web... because I can't run the full suite of ad-blocking software that I run on my computer (Firefox + uBlock, sponsor block; Little Snitch with tons of block lists). I hate that all that ad-blocking software is necessary... but that's a totally different rant.

I _hate_ that I'm being asked to spend $2000 on a device that will, by default, surveil me 24x7. I can spend HOURS trying to disable every feature designed to track me... and still not be certain that somewhere in the TOS that I've still given permission for them to vacuum up SOME piece of information about me.

And I _hate_ that Apple is now an ad company.

And yes, I acknowledge that with respect to privacy, Apple is STILL the less bad choice (compared to Android).

And yes, I acknowledge that there are other options. In fact, when this iPhone dies, I plan on getting a Pixel and putting GrapheneOS on it.

I guess what I'm saying is... I hope that there are more people like me out there... who are sick of paying thousands of dollars for devices that, instead of serving me, serve those who want to collect my information so they can sell me stuff.

edit: "because I can't" instead of "because can't"

> Keep in mind the typical user doesn't care enough to go install any ad blocker

I hate that you are right about this :-)

It really hit home for me today. My project manager was sharing her screen so we could see the free planning poker webapp our team uses. I had it open locally on my machine. She had ads, I did not.

She's not a programmer, but she _is_ technical. She's worked adjacent to programmers for many years, and she's very adapt at writing SQL to query our DB. But she didn't think to install an ad blocker.

I guess I really do live in a bubble where I think _most_ people run ad blockers.

Yeah, keyboard binding consistency is what I really want. Ideally, you could choose between Vim, Emacs, <whatever else> keybindings for the program you're using.

I think what's also needed is a standard way of navigating what I guess I'll call the "focus tree". For example, if I'm running Vim inside of a terminal multiplexer (eg: Zellij) inside of a terminal emulator inside of a tiling window manager (eg: xmonad)...

It would be GREAT if I could use Vi keybindings to navigate windows in whatever "focus layer" I'm at... and then have a "standard" keybinding to navigate up and down that tree (eg: "Ctrl + >" and "Ctrl + <").

Right now, everyone seems to solve the "focus tree" problem by trying to choose a control key combination that won't collide with any other process that might be listening for key chords at the same time.

Many of these aren't animated... which gives me a great(?) idea for a new AI benchmark: "Animate the mechanical movement at this URL: ..."

Way more telling that "draw a pelican on a bicycle"

> The ad system they built isn’t illegal, but it’s pretty clear who it’s for.

"If you're not the customer, you're the product".

Well, Amazon actually did it. They made us the customer _AND_ the product. Hooray.

Like LOTS of other people, I, of course, use Amazon. But, more and more often... if I can find a product somewhere else within, say, 5-6%... I'll buy it that someplace else. Especially if that someplace else is actually a building I can walk into and inspect the thing I'm buying before I pay for it.

Shopping on Amazon is getting worse every day. How can it be good for Amazon in the long run that every time I buy something from their site, I actually have to spend time convincing myself that I'm not getting a knock-off product?

In the modern world, every single time money leaves my bank account I feel like I'm potentially being scammed. I hate it.

Oh, I hope you are right about cars. If one of the outcomes of the memory crunch is that manufacturers start making “dumb cars” again, it would almost make it all worthwhile.

Sorry, knee jerk reaction any time I see “contact sales” instead of a price.

Clicked on the link, ready to buy one. “Contact sales”. Ew. No thanks.

> The US also sees AI dominance as critical for their national security, and relies on a market economy and non-state-owned labs.

[...]

> but the fact the discussion here ignores the fact that these concerns and dynamics are real just lower the discussion level instead of actually discussing potential solutions.

We had this problem solved for DECADES. Build the best post-secondary education system in the world. Lure all of China's (and the rest of the world's) best and brightest. Educate them. Give them well deserved, high paying jobs doing what they love and invite them to build a life and stay here.

Now we appear to be busy dismantling that system.

> When you know you're essentially babysitting the workhorse to ensure it doesn't go off the rails

I don't have data to support this (other than, I guess, my LinkedIn feed), but my impression is that the management class is pushing AI _way_ harder than the worker / craftsperson class.

And if that's true, I think it's perhaps because it's something they understand: you tell AI to do something, and (with varying degrees of success and less complaining)... it does that thing.

To the extent that I've seen craftspeople adopt AI, it's been because they recognize its usefulness as a tool to further their craft. I don't meet many craftspeople that enjoy watching any[one|thing] do their work for them.

I appreciate you stepping in and clarifying. Thank you.

But honestly, this feels like damage control. This is (to me) clearly an innovation on a dark pattern that is basically just "accepted practice" nowadays, namely the "subscribe by default; make it hard to opt-out of said subscription at signup time".

That's why I think people are upset about this. By not taking a credit card, you made it feel like it wasn't an implementation of that dark pattern (yay!) -- however, secretly, it was!

If the intent is to be customer friendly, it's so perfectly clear to me what the answer is. OPT-IN. In other words, a checkbox:

"By default, when your free credits are consumed, all of your runners will be de-provisioned. Instead, if you would prefer that your runners continue working, check this box and we will invoice you for usage in excess of your free credits."

But honestly, you did more than I expect of most service providers today. You sent an email; you actually told them how many credits were left in the free tier. AWS, for example, can only give me an _estimate_ of how much cost I've accrued. They can make no promises about the rate at which I'm expected to accrue new costs. And, unless I've taken great care (by, say, terraforming every resource in a given AWS account), "turning off my cost accruing services" is not a simple matter. If I understood the article correctly, there was, at least, a "single action" they could have taken to immediately stop accruing costs.

Additionally:

If Nvidia is included, does that mean that the money Amazon, Microsoft, and Oracle get for selling compute to the frontier models are included in their revenue?

Because for Amazon in particular, the situation this pages shows is actually much WORSE than I expected. I thought they were making a killing selling compute for model training.

Yeah, my first impression when I saw this was: if this is accurate, the situation is not nearly as bad as I thought.

I do wonder why Nvida is included, though. If you include the company that all of the frontier models are pouring money into, of course the net (expenditure - profits) of the collective is going to be closer to zero :-)

Ah, thanks for the clarification! Would it have been accurate then to have said:

"iff it is bounded and has countable discontinuities"?

Or, are there some uncountable sets which also have Lebesgue measure 0?

> f is Riemann integrable iff it is bounded and continuous almost everywhere.

FWIW, I think this is the same as saying "iff it is bounded and has finite discontinuities". I like that characterization b/c it seems more precise than "almost everywhere", but I've heard both.

I mention that because when I read the first footnote, I thought this was a mistake:

> boundedness alone ensures the subinterval infima and suprema are finite.

But it wasn't. It does, in fact, insure that infima and suprema are finite. It just does NOT ensure that it is Riemann integrable (which, of course the last paragraph in the first section mentions).

Thanks for posting. This was a fun diversion down memory lane whilst having my morning coffee.

If anyone wants a rabbit hole to go down:

Think about why the Dirichlet function [1], which is bounded -- and therefore has upper and lower sums -- is not Riemann integrable (hint: its upper and lower sums don't converge. why?)

Then, if you want to keep going down the rabbit hole, learn how you _can_ integrate it (ie: how you _can_ assign a number to the area it bounds) [2]

[1] One of my favorite functions. It seems its purpose in life is to serve as a counter example. https://en.wikipedia.org/wiki/Dirichlet_function

[2] https://en.wikipedia.org/wiki/Lebesgue_integral

I think it's debatable whether or not it's an ad. I also think it's debatable whether or not the title of the post is sensational.

BUT

It truly warms my heart to see the level of mistrust the comments in this thread show towards (a) venture capital funding and (b) anything even resembling an ad.

> I think most people would interpret “scanning your computer” as breaking out of the confines the browser and gathering information from the computer itself.

That is exactly how I interpreted it, and that is why I clicked the link. When I skimmed the article and realized that wasn't the case, I immediately thought "Ugh, clickbait" and came to the HN comments section.

> To reiterate, at no point am I saying this is good or acceptable. I think there’s a massive privacy problem in the tech industry that needs to be addressed.

100% Agree.

So, in summary: what they are doing is awful. Yes, they are collecting a ton of data about you. But, when you post with a headline that makes me think they are scouring my hard drive for data about me... and I realize that's not the case... your credibility suffers.

Also, I think the article would be better served by pointing out that LinkedIn is BY FAR not the only company doing this...

> People are going to like this-especially the cohort that would buy a cheap TV at Walmart.

What? If anyone truly believes that "People are going to like this", then just make it opt-in.

There is a reason it's not "opt-in". They know damn well people are NOT going to like it.

I'm glad to see this comment and the parent comment voted so near the top. I've had the same experience. In my experience, Kagi used to be great... then it became good... and now it's "better than Google".

"Better than Google" and the fact that I can choose websites to exclude from my search results are two features that I remain willing to pay for, however.

I noticed this years ago with technical books. IIRC, Manning was the first publisher that I noticed doing it. Pages so thin that I could see the text on the reverse side as I was reading it - it drove me crazy. O'Reilly started doing the same.

I had a PDF version of On Lisp (Paul Graham put it on his website for free some time after it went out of print). I used lulu.com to turn it into a printed book (1 copy for myself). I love it. The cover art isn't great (low-res image; not Lulu's fault), but the paper stock is amazing (I got to choose it!). The print quality is also great.

Lulu provides some evidence that you can run a profitable business and still offer users the ability to do _very small_ print runs (1 book). I wish they (or someone like them) could work out a deal with publishers that would let me choose the paper stock I want when I order a book online.

But, maybe there are other options...

Two quotes from the article:

> I purchase most of my books through Amazon. I don’t find the speed of delivery that valuable, but the competitive pricing (especially factoring in Prime), ease of ordering [...]

> To add insult to injury, print-on-demand books seem to be significantly more expensive than stock equivalents

That's the classic enshittification playbook right there. Hook 'em with low prices. Once you've captured the market, lower your costs and raise prices.

Vote with your wallet. Go to a bookstore. Small and local is fun if you don't have a particular book in mind. If you do have a particular book in mind, check Barnes and Noble's website. It will tell you if it's in stock near you. If not, order it. If you go to pick it up and don't like the quality of the print/binding -- return it.

edit: fixed spacing for quoted text

Okay, sure. But if I prefer the subset of CD sound that is the same as vinyl, and my favorite band comes out with a new album... I just buy vinyl, right?

Or are you suggesting that I buy the record, a blank CD, and all of the high quality recording playback equipment I need to write it to that CD?