theneverfox

Infrastructure costs. Their costs don't change with how much data you use, they change with how much data they can throughput

The problem with health insurance as a metaphor is they have real costs... The insurance company does pay out real money every time you use your policy, and that makes it easy to muddy the issue

Let's take the coffee metaphor further. They say "you can drink up to 400ml of coffee, past that we'll add an extra fee. But don't worry, no one does that". Then they refill your coffee without saying a word, they won't tell you how much you've used unless you ask, and they won't stop refilling it unless you tell them not to

The reason the coffee metaphor is great is because, while it's a real thing, it costs them basically nothing. Just like the extra electricity to send your data costs basically nothing

The cost is the number of coffee pots, the labor, the restaurant - all things that don't change in cost no matter how much coffee you drink

Coffee works because the nature of the transaction is the same

Ironically, he picked a metaphor that doesn't support his point at all

If you go to a Starbucks, it's like you're buying a set amount of data. You don't expect unlimited refills, because that's not how the transaction works - you buy the coffee by volume. It's yours with no strings attached

If you go to a restaurant, you buy access to coffee. I do expect unlimited coffee, I would be livid if they charged by the cup. However, you do not get to expect to take any coffee with you - you're using their "infrastructure" to hold your coffee, and you don't get to walk out with the cup. You don't get to share it with the restaurant or the table - you're burying a personal "subscription" to coffee for the duration of your stay

Coffee, like data, is effectively free at a restaurant. They must pay for the infrastructure, but after that each additional pot only costs a few cents. They must make at least 1 pot a day, and a human can't safely drink more than a couple pots in a day (which is an obscene amount only the heaviest caffeine addicts could tolerate). You get it one small cup at a time, if you bought a second cup you could double the rate of coffee delivery... They might even just give it to you for free, because it costs them so little and they want you to come back

You purchase access to coffee for a time, or you purchase coffee by volume. They shouldn't be allowed to charge for both - maybe if you've drank 14 cups and others want coffee, they should be given priority during lunch rush as the rate of coffee production is limited by infrastructure

It's actually a pretty decent metaphor, it just doesn't support his argument at all

Well Google was basically that - it revolutionized search, which made the Internet accessible for casual users

And it worked - Google put more into R&D moon shots than anyone... Except the economic META has changed, and everything innovative just ended up in the Google graveyard before it had a chance to mature

Bell Labs worked because they threw excess piles of money at the best people they could find, and they gave them autonomy. They gave them time, and let them build things with no clear application for their company

Today, that money goes into stock buybacks, executive bonuses, and buying out promising startups. Stock prices this quarter are all that matters, and R&D only raises stock prices when it promises insane growth or quick monetization

Systematic problems require systematic solutions

Because hashes are deterministic one way functions - they're generally one way only

Let's say I hash a picture. It could go from 14MB to 128 digits of base 64 - there's orders of magnitude less information in the hash than in the source data

Now - with that hash can you rebuild the picture? You've lost a great deal of information, you don't necessarily even know the size or the format of the input.

Let's set up an equation - x is the input (the photo), so hash_func(x) = hashx

There are multiple, maybe infinite (depending on the hashing function) values of x that will solve our equation. In the case of the photo, most of it will be random combinations of pixels that mean nothing to a human. There could also randomly be things that appear meaningful, but without knowing more about the original you could never be sure if you have the correct answer

Now, passwords might actually be shorter than the resulting hash, but we salt them so each password hash function works differently, and can still destroy information from the original password. Part of the password and the salt are then used as basically the seed for a deterministic random function to generate this extra information

Again, you have the dual problem of a huge problem space as well as an inability to be sure you have the original input or just another solution

Ultimately, everything is defeatable, and if you can narrow down the problem space (say, by knowing the length of a password, having enough known before and after data, or finding a bias in the algorithm), you can reduce the needed computations by orders of magnitude and make it feasible. Quantum computers also grow exponentially with chained qbits, so I expect someone clever will figure it out sooner or later

...Do you not think corn can be turned into a carcinogen?

... So a short form video platform is more dangerous than a dubitably secure group chat app?

That's comparing apricots to ocelots

There's different types of hacking. Finding and coding up an exploit? That takes skill

Then, they post it in a corner of the Internet somewhere to get appreciation for their achievement, or maybe even sell it on the dark web (or someone else sees it and packages it up in a state to sell/share)

Now, using the exploit? That's pretty easy. It requires some technical ability, but not much. It's just installing and configuring stuff, then using an app

So here's what I think happened. Someone found the exploit, and posted about it in a hacker community off the beaten path. One thing led to another, and somehow a group of edgelords get a hold of it. As a group, they manage to get it working, and act like edgelords

The original hacker might have been related, but real hackers are cautious or quickly caught - they probably solved the puzzle, maybe played with it a bit, then posted their findings and moved on to the next puzzle

The best tool is the best tool for the job. My php code was immaculate - import header, write your content, import footer. I wrote it all in a few hours, and I think my university is still using it today. One of my proudest moments was when a team member whose abilities I respected gushed over how easy my code was to work with

I haven't used it since.

I'm led to believe modern PHP has become pretty good. It's also never been the right tool for the job since then

Mojo sounds incredible - python is a beautifully consistent language, and I've written a shit load of it. I like it a lot. It's often not the right tool for the job, but mojo promises to turn it into something extraordinary that shores up its weaknesses... It's just promises so far though, I really hope they pan out

But always ask yourself - what's the right tool for the job?

So what could be better than getting started?

Yeah, if search hadn't become dog shit I'd be happy with it

Instead, everything is a video for some reason, and the results are purposely worse than a year ago...I don't want to watch a video, I can read 20x faster than I can listen, I don't want to read an ad in article form - I'm generally looking for one little nugget of information

I took this into my own hands - I'll use free services if they work, but increasingly they're just demos for a product that may or may not be better. So I spun up a searx container, I point a local LLM at it, and I let it filter read through results. My next stage is to crawl documentation, use LLMs to feed it into a vector db, and use AI to retrieve exactly what I want without sifting through garbage myself