joshspankit

Super cool and seems to have a depth of historical knowledge

I’m shocked to see that LORD can’t be used because of licensing. Legally I know that’s how it is, but it Just Feels Wrong(tm) that a popular game based on 9600baud phone modems is encumbered in the era of fibre, wifi, and smartphones

Indeed. Vital to the solution yes, but if it was all that was needed then the surveillance systems would have been dismantled and prevented each time enough people became aware. The fact that we are here now is it’s own evidence

I’d even say that YouTube autoplay is “watching TV”

You’ve missed the point: each addition can (and likely will, based on history) be the way to bring more patents into the market. It rarely has anything to do with objective quality (except in the marketing)

Any chance you’ll be adding diarization of known speakers?

For example: from audio samples tagged with names

Through all this I kept wondering (and excuse the question if it is inappropriate): what’s the benefit of a transcription of the sounds since the ideal is a 1:1 IPA representation to audio recording and neither is itself a translation?

Isn’t the recording itself “more ideal” than any representation of it?

You may be underestimating the desire to earn through differentiation.

Assuming the product companies accept the plateau on resolution (I agree it’s likely), there are many ways to exploit the other stuff. Colour space, tacling each type of compression artifact, additional channels (probly not long until we have HDR RGBW), etc etc and those will each come with their committees made of patent-hungry companies seeing non-encumbered encodings as “undesirable”

Definitely curious about the breakdown of data-broker relationship for each link in the chain. Unencrypted data is easy to store and share so it’s reasonable to assume that even the most stringent privacy policy for an IoT device is essentially meaningless if it’s unencrypted PII passes through a single tracking server (including the user’s ISP who may have a direct relationship with multiple brokers)

This has always been true to some extent: some have used weaponized security tools against competitors. It comes with legal risks (as it does with AI) but overall it doesn’t really improve the security baseline unless those holes are made public

It’s systematic

“6-12 mo to shaky profitability + ability to quickly iterate” is a business that has a good chance of surviving while “However long it takes to be fully secure” is a business that is not only rigid but needs massive up-front capital to get there and even then there’s no guarantee that the market fit is right

And after that is something we could call the “Pareto spiral”: if a company find market fit and builds an excellent product, competitors can survive at 80% of that quality. If the “100%” fails for any reason, the competitors become the new ceiling and now their competitors can survive at 80% of that (now 64%)

And only one round in, how secure could that 64% company be?

The current reigning social script is that drone enthusiasts are piloting all the drones. I know this will change as articles like this come out but for now the police are in a very convenient bubble of essentially undetectable surveillance

Palantir helps everyone fight everyone

As someone with a long-time interest in Tim Ferriss I’d say he’ll probably go most places he’s invited out of pure curiosity. He’s also attended TED and has been in many VC spaces so it’s not surprising that someone (maybe even Thiel himself) would invite him

I’d like some confirmation if possible, but my gut says the general internet (and therefore the training data) probably has a lot of sloppy Python code along with a lot of sloppy writing about Python whereas Phoenix has majority good code and well-reasoned writing

What I mean can be shown with an example:

Let’s say first that we know (some) users will inevitably agree to let malware compromise their system, no matter the popup or protections

A compromised system that’s transparent:

- Has only one way an executable can be started and, being designed as a “salt flat”, it’s easy to read

- Exposes all I/O and all network requests (to admins), regardless of driver abstractions

In this case, even a young enthusiast can look at a system and immediately see that it’s compromised, remove it’s ability to start or do work, and likely remove it from the system entirely.

The inspiration for this approach is a backlash against the absolute glut of places to hide in current user-focused systems. From multiple startup options, to services, to drivers, and in to the “hidden from the admin” executables that can be compromised it’s an ever-worsening problem that erodes user’s ability to keep their own system secure

For a long time, I’ve believed that the actual solution is to make the system transparent enough that a compromised system is obvious. Imagine playing hide and go seek in the salt flats

Downvoters: I’m curious why

Devs and other people who have seen behind the scenes at large companies know that most security is at best shaky and mostly hand-waved

It’s not even really the fault of the people who pushed for these setups, it’s a seemingly simple business decision: build it in a way that looks secure, add some black-box process, and tell the overseers that the reason there are no attacks is because it’s bulletproof, and definitely not because no one has really tried

Then, when someone finally turns their attention to you and walks in: fire whoever needs to be fired, patch that specific hole, maybe spend a bunch of money on a different system, assure the overseers that it’s handled, and move on with business as usual

It’s cheaper in the long-run, it makes stockholders happy, it relieves the bosses and their bosses, and for the most part there are “no security holes”.

Until now, of course

You attempted to create a deterministic test for an N-dimensional non-deterministic output

I also suspect the intense vibrations have a similar effect…

My gut says that there are some interesting discoveries waiting around the intersections of frequency of vibration, individual resonant frequency, and duration

Someone really needs to properly do the science on this

I (presumably like the majority) assumed that sleep apnea was at least partially caused by weight gain, but if there is weight gain caused by sleep apnea it’s going to give doctors some new tools

Underrated question

Personally, when I have not slept well and need to be productive in a day, I’m much more likely to want to load up on sugar and unhealthy food

It's not even quite clear what the article argues against.

I think it can be summed up as “average CPU utilization, which is the common and intuitive first check doesn’t tell you the real story”

I would also suggest that these are “outdated” measurements as common CPU metrics are really designed for moderately multi-threaded, single-foreground-application on bare metal

To your point, someone who deeply understands the stack already knows these are not the metrics to look at, but this is clearly aimed at people who have not (yet) had to dive deep to figure out a scheduling issue

You might be interested to learn exactly how fast some listen to screen readers, and how many overlapping sounds can be successfully navigated

It seems like the right solution is around the corner: placeholders for these kinds of strings (uuid, hash, etc)

Why should an LLM even have these types of IDs anywhere in the prediction pipeline?

Google search results have been the worst part of every LLM I’ve used. I imagine the LLM specifically designed to use Google search is going to be the worst LLM.

Anthropic has the private key now too

edit: Personally I don’t think they would take advantage of it, but still worth moving the BTC asap

Yes but

I think you’re underestimating “average users”. If we talk about the median, then probably you’re right, but if we talk about “the group of people clustered around the average” I think there’s a lot of untapped potential, especially in people who assumed data and programming were unknowable/impossible and have therefore been held back by “good” tools like excel

A bit of UI/UX sand: Store the scroll position when navigating the list.

User clicks on a project, views the details and returns to the same position in the list (including all the expansions via “Show more”)

What if we instead look at the last clip as “the one shot”

Personally I don’t think that takes anything away