Everyone has some economic game going on. If some entity can see most of the cards you hold, it like putting your cards open on the table during a poker game. That is why big companies want your data, they want to peek at the cards of as much players in the game as possible.
Information asymmetry has always been a thing, wars have been though over this.
But I think that in our age, information asymmetry is particularly low, at least in western countries. Each one of us has access to a tremendous amount of data, sure the powerful have access to more, but I have a feeling that the relative difference is shrinking.
I will always remember when a police investigator was interviewed, the context was a controversy about police files. The investigator said: "police files? not very useful, when we want to investigate someone, we browse Facebook". It means that the police doesn't have much as much of an information advantage compared to you and me.
Journalism, world events, etc... Most of the times, we have all sorts of first hand reports, photos, videos, news sources from enemy countries, etc... Not all of them reliable, and factchecking enough to see through that mess takes work, but it is possible in a way that wasn't before. A lot is available on open data platforms, plus all the shady stuff like Wikileaks, darknets, etc... that are not that hard to access either.
Should you want to, you can be your own Palantir, because most of what Palantir does is standard data analysis that can be done with open source tools, and most of the data sources are public, private data is just the cherry on top.
Of course it takes work, but it is possible with limited resources, mostly a computer, an internet connection, and time. No need to travel around the world to meet contacts and get access to paper archives.
A simple WIFI/bluetooth only device like the iPod Touch but with Linux, combined with a modem puck would actually be enough. You separate the untrusted part from your own device.
Well, just running on a 6C/12T Coffee Lake CPU, (I'm looking through these speeds in LM Studio as I type this..) I got like 2 tokens a second with Deepseek R1 14B, 3.4 with 7B Qwen, and 4.4 with 8B Llama, although out of those two I found 7B Qwen's answer to be a bit better. (My GTX1650 has 4GB VRAM, loading 1/4 the layers is pretty ineffective, GPU util went up to 10% and I gained like 1 token a second LOL.)
So it'd take a minute or two to type out one of those answers where it's got about 4 or 5 beefy paragraphs of thought and a decent sized paragraph for it's answer. I'll put it this way, I can type 120 WPM and it puts out text a bit faster than I could write it.
Input's a LOT faster though, I was asking these models to analyze a document so my input was like 2200 tokens, they all did well over 100 tokens a second on input.
I was playing with a self-hosted model a while back and instructed it to only give answers that were unhelpful, vague, and borderline rude.
It worked surprisingly well a lot of the time! But most of the time it also kinda broke the model in terms of coherent answers because it was obviously trained for the exact opposite thing.
At university I made some money on the side giving courses in HTML, Excel etc. One person that was mandated by an employer to follow such a course had extreme trouble using the mouse. I opened Solitaire and showed quickly how to click and drag. The person in question had to confess he also did not know how to play Solitaire.
When you don't want OpenAI use your ChatGPT chats for their AI models by turning off "allow them to be used to improve our models", it is no longer possible to have longer conversation then a few minutes.
When you focus back on the tab with ChatGPT after a few minutes, then this event is probably used to wipe the conversation, because you can see the page getting cleared when you want to continue.
Before this issue it, was possible to have a conversation open and stable for 6 hours (already not ideal).
This new behavior hinders usage like solving software development or system admin issues. This also affect users with the Plus subscription that might want to use ChatGPT in a professional context and can't offer company data to be used in OpenAI models.
It is not clear if it is an honest mistake or a newly introduced limitation to nudge people into sharing more data.
I have set up two NUCs with PopOS for my sons. Both very happy and I have full control of the machines.
The problems started with Windows 7 when at some points the updates became obfuscated and mandatory to push "telemetry" aka data snooping. Before that phoning home was considered spyware and a sin.
While researching hardware for solar panels I scanned the web for possible hardware and saw that Enphase is supposed to have good inverters needed create AC. I am looking for hardware that does not need to be connected to the internet for all the obvious reasons like security, privacy, continuity, stability, resale value, etc.
So I asked myself the question; can Enphase hardware run without interaction with the Enphase company. Someone in their forum already asked this question:
What I understand is that people are forced now to periodically get a token with limited lifespan from the Enphase website to access the API of their hardware in stead of open local access. This seems to upset people with 20.000$ setups.
I am curious if I am reading this correctly, because this looks a bit like a "Bait and Switch" in the legal sense (not a lawyer btw) if this is true. It seems that Enphase removed a well known and coveted feature (independent local api connectivity) after the sale for the benefit of themselves. I am wondering if people have experience with this hardware and the situation, because I might be wrong and don't want to wrongfully suggest an issue. Hardware looks nice, but if this is true and does not get reverted, it would be a deal breaker for me.
reply