Slow is smooth, smooth is fast

Multiple well-established twitter accounts appear to have been hacked and have sent me phishing links asking to vote for them to be on some podcast (link usually has “.top” suffix and asks you to “log in” with your twitter or insta account). Now is a good time to get serious about digital security if you haven’t already. Worth an afternoon (*extremely* tedious but put on some cool music/sunglasses and pretend you’re Neo): 1. Get 1Password 1Password.com 2. Put every single one of your logins on it, and make them good unique passwords. All your API keys too 3. Every single password has either 2FA or passkey as extra security. 1Password itself can be used to fetch temporary codes (more convenient than having separate app like Duo Mobile) 4. If your agents need a password/key, put it in a separate 1pass vault for them to use 5. Make sure all sensitive computers have good protection from remote login, e.g. via tailscale, so only your physically secured devices can reach them 6. Make sure your agents know not to put keys or passwords in plaintext, and routinely scan for this on your computers Once this is set up it’s quite easy to maintain (just two standard rituals: new login means add to 1pass and add 2FA, new machine means add to tailscale network) and LLMs can help a lot with setup I think this covers 99% of vulnerabilities, at least at an individual user level

Stray thought: the number of ways to combine n things scales inconceivably faster than n (more possible QR codes than atoms in the universe), so systems of combining things in meaningfully distinct ways—words, concepts, amino acids, social groups, etc—quickly become in a certain sense bigger than the universe (“the brain is wider than the sky”) And arguably these principles of combination are the only reason anything interesting or new happens in the universe at all. If things simply aggregated in meaningless or indistinguishable ways (e.g. one pile of sand isn’t that different from any other) then nothing truly new could ever exist, or happen So in a way, the story of the universe—of interesting *new* things happening—didn’t begin in earnest until biological life began, specifically when the genetic code for building proteins became operational: the first process for constructing arbitrary combinations of things with emergent properties and causal powers. Intelligence and social living provided further such processes Looking at things this way inverts the Copernican picture and puts earth back at the center: the physical size of the universe dwarfs us, but nearly all interesting *combinatorial* space is occupied by things happening on earth, with the whole rest of the inanimate universe just a slim margin It is a cliche to say that what happens on earth doesn’t truly “matter,” because we are so small. But if we are going to decide what “matters” based on magnitudes (already dubious), why not tally based on how many *unique* kinds of things there are? If there are just 100 interestingly distinct categories of stars, it is not obvious why a quadrillion ~copies of each, arranged very far apart, should “matter” any more than the first 100. We are just a pale blue dot but that is probably where most of the distinct kinds of things in the universe happen And it does not take very much to get bigger than the universe. Emily Dickinson’s poem is 72 words, and I am told by Claude that between 10^100 - 10^200 meaningful sequences of 72 English words are possible, vs 10^80 atoms in existence. Just by expressing a new thought and choosing your words with care you have navigated a space bigger than the universe (a reason not to surrender that power to LLMs, as neat as they are—why allow next-word prediction to fence in that space for you?). So not only is the brain wider than the sky, in a way so was Dickinson’s poem expressing that very thought I think these are decent reasons for a human-centered (or at least life-centered) worldview

all the twitter folks who’ve been going back and forth about whether AI could be conscious should check this out: in BBS someone writes a target article, many authors respond (often experts in their area), then the original author responds to the responses. One of the best ways to get up to speed on controversies about brain-related topics