Just make a route on your web server, making all the files available with some long, impossible to guess, unique ID that can be shared. Like https://files.<your domain>/<id here>.
If they want to collaborate, they can just post the changed file, using the auth key you generated for them set in some header field, to https://files.<your domain>/<id here>, which could automatically increment revision numbers. Then you could access specific revisions with .../<id here>/rev/<revision>.
So much easier than installing an app! You could literally just use curl as the interface! (I kid)
Every so often someone is like, Dropbox isn’t that hard. Look at this amazing ZFS/whatever! So simple. Yeah, I keep paying Dropbox every year so I don’t have to think about it. I shoot a sync off to backblaze every once in a while.
I dislike Dropbox for reasons that aren't technical, but the big thing for me is that I want either E2EE, or control/ownership of where my data is stored. These are my personal files (no, not that kind of personal), I'm not just going to scatter them on the internet.
My solution so far has been NextCloud, but I'm getting pretty fed up with it. But not enough to actually do anything about it... yet.
> I dislike Dropbox for reasons that aren't technical, but the big thing for me is that I want either E2EE, or control/ownership of where my data is stored.
You could run something like Cryptomator on top of Dropbox:
Boxcryptor used to be amazing for this till they got acquired by Dropbox. Last I checked cryptomator couple years ago it still hand random file corruption risk, is that still true?
I do agree with you at a philosophical level. I have worked in infosec long enough to know. I am pretty careful with what I upload. It’s just hard. Every little home hosted thing. They eat your time. Take effort. Even the “easiest” solutions have a real human cost if you are hosting it yourself.
My solution is… I have no fucks left to give about it. I haven’t for a long time. It works. My family will have all our photos and valuable sentimental data preserved. I keep a local backup. I spend my time on other stuff more valuable to me. If dropbox and backblaze disappear tomorrow oh well. If all my data is leaked? Knock yourself out. All the good stuff is in encrypted volumes.
My data has been breached 12 times by my count. /Twice/ by the OPM itself as I had a security clearance. DOGE goons have all our data and walked out with USB drives with all of it.
Equifax and credit agencies are a joke. My threat model is simple, nothing you can blackmail me with goes to the cloud. And that’s that. Breach me, try to hack my finances, try to steal my crypto. I have had my SIM transferred and someone unsuccessfully attacked my crypto accounts with it.
I love Dropbox, I pay annually. I use the open source client https://maestral.app/ on the Mac for workstation use, but also integrate other systems with my Dropbox account using their API. If someone built an open source Dropbox server that sat on top of S3 compatible storage, I would not only use it, but pay to have that optionality to get out of Dropbox if they ever enshittify. I can recognize form and function worth paying for ("value"), but still want an exit plan. It's not about the spend, it is about data sovereignty. This is colloquially referred to as “vendor and third party risk management.”
The Manhattan project employed some significant % of all of America. A project of that scale will likely never happen again.
It was also about far more than the science. It was about industrializing the entire production process and creating industrial capability that simply did not exist before.
I am skeptical you could do something of that scale for 30B today. That is just the dollar cost based on inflation. If you used CPI indexing probably hundreds of billions to a trillion dollars now.
Does quantum computing need that though? We don't suddenly need a large, unique supply chain for these computers. We don't need to dig up the qubits and refine them. Testing doesn't blow up the computer.
The Manhattan project had a huge impact but it was not that big as far as efforts in the war went (they managed to hide the budget allocated to the project from most of congress, for example).
It was like 0.5% GDP. It wasn’t insane, but still, you are right.
It was very focused. What do we get out of the F-35 program? By comparison, it has eaten (projected total lifetime cost) 2 trillion dollars. It is 4.5% of the GDP. I had no idea. This is just a military and government contractor subsidy. What are we doing…
Mortgage loans and car loans in America also ask for your W-2 or proof of that said income. Can't prove it, they won't let you use your claimed income as basis for loan approvals.
I've even had to prove my salary when applying for apartments. No loan involved. The first time that happened I didn't have a w-2 yet, so they called my employer to check.
I never had to supply a W-2 for a vehicle loan. Just my SSN. I almost always get unsecured loans too. They just dump money into your bank account and say "please buy a car with this" and you keep the title. For a mortgage you do have to validate your income. The work around is to just buy your house in cash, I guess (sigh). So I guess mortgage loans leave you open there, but those happen less frequently and may not be good map of income level on a shorter period.
Once in a while they want to see a bank statement or two showing actually paycheck deposits, but I only ever saw that on a mortgage. Once for the car loan they asked to see a balance or two via bank statement. So I showed them a bank account sitting around the $$ for the vehicle loan.
I tend to just avoid loans if at all possible now though.
An unsecured loan just a straight cash loan. In my experience the interest rates are usually a fair bit higher than a vehicle loan, and I would assume the maximum amount is generally quite a bit lower.
In my experience it really depends. You are still contractually and legally obligated to purchase a vehicle in many of these setups, they just don't follow up or enforce it directly in most cases. Loan rates are nothing like unsecured personal loan rates. Probably higher than best auto loan rates. With rates being what they are it may not make as much sense to use an unsecured loan at this point, but when rates are low the cost was negligible and the convenience worth it.
In Maryland I received my car's title, but it just shows the lien holder's information on it. I've never bought a car in any other state so my information is limited.
i'm pretty sure that in Arizona, and i think in California, they DMV sends the title updated to show the original holder (like BMW Financial Services, Subaru Financial Services, or whatever) has been payed completely, and the person is then shown as the title holder.
Congrats. For those of us who can’t afford to build one, we still enjoy the heat through other means.
Or at least I do.
There’s just something extraordinarily relaxing about going from the high heat (though obviously not too high) until one can’t bear it, then transitioning to a cool off.
It’s nice. Sauna doesn’t have to be that expensive. On par with a hot tub. Way less maintenance. Can be installed outdoors. It does suck if you can’t. Public Sauna run at a nice temp are not easy to find in the US.
This helps, but the original prompt is still there. The system prompt is still influencing these thinking blocks. They just don’t end up clogging up your context. The system prompt sits at the very top of the context hierarchy. Even with isolated "thinking" blocks, the reasoning tokens are still autoregressively conditioned on the system instructions. If the system prompt forces "caveman speak" the model's attention mechanisms are immediately biased toward simpler, less coherent latent spaces. You are handicapping the vocabulary and syntax it uses inside its own thinking process, which directly throttles its ability to execute high-level logic.
I get your point but it seems that extended thinking is based on a hidden system prompt that is not so much affected by the style the user defines. Probably it's a bit in between.
That is part of it. They are also trained to think in very well mapped areas of their model. All the RHLF, etc. tuned on their CoT and user feedback of responses.
reply