Thespian2

Personal AI LLM chatbot, trained on my emails and slack messages. I will have succeeded in replacing myself with a small shell script.

Extraordinary claims require extraordinary evidence.

Open project (code available - results to be made available) to gather weightings on newest Stable Diffusion outputs, for which the audience likes best. Accepts filtered prompts. Requires Google or Discord login

If this is a continuous function, that implies there is some time horizon between short-term and long-term where our ability estimates are spot-on.

Finding that precise time horizon is an exercise left for the reader.

Story seed ideas for improv role playing games, or quick, off-the-cuff stories for kids.

Whatever it was seems to be down

In the future we will no longer "hack" computers, but rather, we will social engineer their front-line AI interfaces with pretexts.

We may not like it, but your last point is almost the exact definition of what a "price" is. Things don't have objective "prices" as an intrinsic quality. They don't even have "reasonable" prices intrinsically. A price is no more, and no less than an agreement between buyer and seller to make a transaction happen.

"how much you can be fleeced into paying" while a proactive way to phrase it, is also "What you are willing to pay." Key words - "you are willing". If it's too high, then it isn't a price, as there is no agreement.

Seller wants highest price, and buyer wants lowest.

When it's a commodity, like apples at the store, with many sellers of a basically undifferentiated product, prices average out to something we think of as "fair." But when a product is unique, or there is a monopoly on it, seller has a huge advantage in pricing.

Exactly. Need to see DMARC, SPF headers. Email headers are what you want to see. "From: <blah>" without those means nothing, and is trivially spoof-able.

I don't see headers in the twitter post?

Same. I've spent a number of nights trying variations on the install instructions. Got the windows CUDA drivers (older graphics card), but no matter what I try, conda, or WSL, pytorch refuses to see the CUDA available.

When we started dating (over 20 years ago) my now wife and I had "yours, mine, and ours" piles of money. Our salaries went into our own accounts, each month we contributed a chunk to "ours" and paid for combined stuff from there.

Once we got married, we became a single financial entity. It's all "ours". Now I'm the salary person and she does way more labor than I do being the primary kid-care person. We help and support each other, and the family, together.

If you really are committed partners, let the ego go. Accounting who earned what may have a place in early relationships, but as others have said, is just exhausting.

ISO-8601 for thew win. Accept no substitutes.

But yes, "act with empathy and gentle guidance" is The Way.

Also, never name you previous versions foo.new and foo.old or you will end up with foo.old.old.no-really-old, and foo.new2

Be lucky.

I don't mean to be flippant, but given the constraints, your speed will be almost entirely proportional to luck/risk.

To start, all you have is labor to bootstrap with. So let's say for round numbers $10 / hour, at 40 hours per week is $400. or $1600 for one month. That puts you six "doublings" away from $100K after one month.

Go to Vegas, bet it all on black (or red, your choice) and win 6 times in a row, and done. 1 month of labor + a few minutes of very lucky gambling. Or, a single bet that pays off 62.5 to 1. While the odds are the same it probably seems more "practical" to win "just" two 8:1 bets back to back.

Investments, options tradings, sports betting, etc will all be some variation of this. Get really lucky, and it can be done in a short time, but with a high risk that you have to start over, and it ends up taking a longer time overall.

More realistically, if OP can put all that labor money into investment/savings (assume subsistence is supported otherwise) The standard advice to buy and hold low/no- fee index funds could do it in about 5 years (on average).

https://www.investor.gov/financial-tools-calculators/calcula...

400 initial, $1600 / month, 5 years, assume 7% yearly growth +/- 2% variance.

Finally, OP says "zero financial investment" - but what other resources do they have? Perhaps the sunk cost of education or training or other skills. Then, their labor might be worth $100 / hour, instead of $10. Get a high-paying job.

Then you could make it in 6 months of straight labor income without risk of investment or gambling losses.

All the above is wildly speculative, and not investment advice, and somewhat US-centric, and contains completely implausible numbers and assumptions, but is an honest attempt to quantify an answer to the question.

Sounds like the most likely common element is phone number. You could have been the victim of a SIM swap attack, where the bad actor used your number to do BadStuff(TM) getting your number on a block list. Or, if you got the number recently, it could have a bad reputation attached to it.

I hadn't used gigablast before, but a quick test had it find some very old, obscure stuff, as the top hit. Well done. However, the link on the front page to explain privacy.sh comes up with "Not Private" in Chrome. The root Cisco Umbrella CA cert isn't trusted. Oops.

That would be a "line feed" (0x0A) - Carriage return is (0x0D) (ascii values 10 and 13, respectively) the Mac/Windows conversion is the difference between CR and CR/LF

All I can guess is that in terminal, when that particular version of bash and OS interacted, it didn't go back to the start of line, and the CR was unprintable, so it all ended up on the same line.

I'd expect to have seen something like what you describe with the line break, if it were a bare LF, but in my case, it wasa bare CR.

Reminds me of copy-paste shell script error, between windows and mac, where first line of shell script turned out to be

#!/bin/bash^M

That is, a non-printing carriage return after "bash"

and running the shell script "helpfully" kept telling me: "/bin/bash: no such file or directory"

As of mid-day wednesday

Chrome blocking site as "unsafe" due to certificate error. Over-riding that and OpenDNS blocks it as malware.

Creators may want to investigate.

SMS is better than no 2FA at all, but is the weakest choice. Avoid if possible, unless it is your only choice, in which case, it's still better than nothing.

Authenticator app, hardware token (e.g. Yubikey), or email will usually be a question of which ones are supported by the service you are 2FA-ing into, You usually don't get a choice, so use what they offer.

Phone-based 2FA and lack of network is only a problem if you are using SMS push, and have no data service on the phone. But in that case, what system are you using to access the service? Can that machine hot-spot, or can your phone share its network connection? In practice, I've never seen a case where you have network connectivity to talk to a service, but not phone connectivity to get your SMS 2FA. So I'd say "not a frequent problem".

Given my choice (assuming all are supported) my order of preference would be: 1) Hardware token (e.g. Yubikey), 2) Authenticator App, 3) email, 4) SMS

Tokens are portable, can move between machines, have nothing to remember, and are easy to use.

Apps are running on some device, so still a true second factor (something you have) - and again, have nothing to remember, and can be secured with biometrics.

Email - now you are on two different things you know (not really a second factor), and you have to access your email on some potentially untrusted device, just to get the code, potentially now compromising your email. It's one more basket that if someone gets access to, they can probably reset your other service passwords for total account takeover. Still better than no 2FA, but higher friction to use and access, and not really a second factor.

SMS - it's just too easy for SIM swap attacks, and good social engineering to phish SMS codes. "Hi, this is $BANKNAME - we saw some suspicious activity on your account, so we are going to send a one-time code to you now. Please enter it back to us here to verify your identity" (scammer then uses your stolen password, tries to log in, real bank sends you real code, which you now are tricked into sending to scammer to complete their login as you).

Back of the envelope calculation. Say you are really tall, 2 meters. That's "only" 2000 millimeters. So even additively, that's still only 0.00000000000002% difference between your head and feet. Let's say our two meter giant lives 100 years. That's a bit over 3 billion seconds, and for their entire life, their head and feet continue to age differently at this rate. If I haven't dropped a decimal point somewhere, that's still only a life time difference of less than a microsecond (6x10^-7).

So "adds up quickly" is relative (heh) in this case, and inconsequential in absolute terms, at human scale.

While I agree, there still remains the practical question of who would pay. In the physical book library world, the library pays for a quantity of physical books, and lends them for free to the patrons. They are reimbursed for various fixed costs by existing grants, and taxes, and similar funding streams. The physical scarcity supports this model, by capping costs.

If we remove the scarcity, and pay authors per reader or per page (a good thing, in my opinion) we uncap the expense the library must bear if something becomes popular. One could shift the costs to the users, but then what is the point of the library? The user is just buying an ebook, and we already have markets for that.

The library gives access to books to those who can't afford them. Artificial scarcity sucks, but actually serves a purpose in this context.

Free tier on Amazon EC2 instance?

If not, I've had good luck with Linode in the past. No longer a current user, but was happy with their service when I was.

https://www.linode.com/ https://aws.amazon.com/free/

Their second description is not 2FA, but password recovery.

At best, the support rep is confused between the two, very different, flows.

Your understanding of 2FA is correct. If what the support rep said is accurate, I wouldn't trust them with my money.

You mention storing the matrix in memory and recomputing hourly because of the small data set. What are your estimates for scaling ceiling with current implementation? How many users / data-items (order of magnitude) do you think you can handle with current set-up?

Many jobs can't be done "remotely." What about food service, cooks and waiters? Delivery people? Doctors, surgeons?

Randel Munroe of XKCD fame walks through this scenario in his "What If" book.

Link to sample on books.google.com

https://books.google.com/books?id=tgZIBAAAQBAJ&printsec=fron...