Thank you for saying this. I feel like your car hack comes down on the effort side of the equation (and I admire the hack!), and I also want to say that I respect what you're saying about memory and motivation. I'm struggling with my own limits and digging into this for those reasons...
B.J. Fogg makes this point in his book Tiny Habits. And his point *is* that you can form a habit in much less than 21 days, let alone 66.
This whole question revolves around the effort/reward ratio of a behavior. When people talk about ~21 days, they're talking about doing a hard thing until it's second nature and seems easy.
But there are other ways to make something seem easy, and there is another component in the ratio: reward. That is, even if effort stays the same, you can wire a habit by making the behavior more rewarding. (This is why people are able to get addicted to a substance after one dose -- because they can't forget the state they entered ... and it was so easy to get there.)
So the takeaway here is the you can wire habits by decreasing the amount of effort to do something that you think is good for you -- eg if you want to hydrate more, place a glass near the sink so you drink water when you get out of bed in the morning -- *and* by increasing the reward. The whole trick is getting the ratio right.
Cliche Silicon Valley example. I did an ice plunge, and it gave me a day long plunger's high. I didn't need to plunge for 21 days to get the habit. I started doing it 3 times a week after that, because I knew what I had to do to feel good.
This actually gets to something Huberman calls "duration-path-outcome". Getting clarity on what you have to do (path); how long it will take (duration); and what the payoff is (outcome), can do wonders for motivation. Confusion kills action (and for that matter, all deals, since habits are just deals we make with ourselves). If you can get clarity, reduce the effort, and increase the amount of reward and your confidence in it, I think you can get to new habits really quickly.
One of the main advantages Anthropic currently has over Google is the tooling that comes with Claude Code. It may not generate better code, and it has a lower complexity ceiling, but it can automatically find and search files, and figure out how to fix a syntax error fast.
A better way to think of it is: you need to live in a host state that may be hostile to you (refugees fleeing conflict or natural catastrophe). You need to work and can do it remotely. A network state might possibly facilitate your work and payments regardless of the host state's laws.
We're in a Cold War with hot proxies. The original Cold War was also like that.
Now, China, Iran and North Korea are backing Russia without engaging in direct confrontation in Ukraine; the EU, NATO, and maybe still the US are backing Ukraine. If Trump realigns with Putin against Ukraine, then the axis of authoritarian countries will have won a major battle. Trump will ofc lose many of his old-guard GOP supporters if he does that.
personal theory: burnout is basically when you're expending more energy than than you're able to recover during rest -- over chronic timelines. that is, i think getting physical health right addresses many causes of burnout. getting your device and app use right addresses a set of others. finding ways to connect with your team, users, company mission addresses other potential problems; ie feeling alienated and on a futile or harmful mission is a great way to get burned out.
ultimately, maintaining focus is tied to overall physical energy (because suppressing distractions consumes neurological energy), and that physical energy is dependent on both health and ultradian cycles. best to alternate between focus bouts and recovery periods.
BJ Fogg's book Tiny Habits is great, particularly about the important of motivation vs ability in starting new routines. His main point is that ability relative to the task is much more important than motivation, because motivation is volatile. That is, it's much more likely that your ability to do something will remain stable long term than your motivation to do something. So if you choose really easy things to do to start a new habit (drink a glass of water in the morning), your motivation won't matter much, and the habit will stick.
So I think Piers is skating over that very important part of the equation.
Another piece of it is, if you manage to get healthy somehow, sleep better etc, then your motivation changes; the base level is reset. You have more energy to spare which you can devote to goals.
But I think there is a simpler way to think about motivation, which comes down to the ratio of effort to reward. The smaller the effort-reward ratio of a given activity, the more likely one is to do it. That single idea seems to rule my own behavior and that of many people I see. But it's also something you can hack, partly by using Fogg's idea of starting small. To change a behavior (and ultimately, your life), you just need to find a small enough starting activity to trigger action. It's not about motivation at all, and all the "motivational speakers" out there are misleading people in some fundamental way about the path to change.
One of the traps in that dynamic is that as we decrease the magnitude of the activity (drink a glass of water as oppposed to "go to the gym once a week to get stacked"), our motivation decreases as it loses its grandiose visions of change. I don't think task size and motivation necessarily decrease at the same rate tho. And I do think that grandiose visions are sometimes a form of self-sabotage or psychological homeostasis; ie "i'm only motivated to do things that i can't follow through on."
Sincere question - why doesn't RL-based fine-tuning on top of LLMs solve this or at least push accuracy above a minimum acceptable threshhold in many use cases? OAI has a team doing this for enterprise clients. Several startups rolling out of current YC batch are doing versions of this.
Some authors of this piece are deeply involved themselves in building robots and other hardware technology. They should be taken seriously. The US is moving into a trade war, however unwise, with the country that supplies components for everything we make and need. That country has expansionist ambitions and a superior manufacturing base, which is typically what wins wars.
Please keep in mind that Dan Hendrycks helped write the disastrous AI quashing bill SB 1047, which Newsom vetoed last year. If these people get their way, the US has no competitive AI strategy at all. He has moved on to pretending he's not a doomer. Nothing could be further from the truth. During his time at Cal, Dan was telling people to get their hazmat suits ready for the AI apocalypse. These are deeply unserious people whose work will have serious consequences if adopted by those in power.
This book was truly excellent. Shows the precursor network triggering similar benefits and harms to its successors. Also a great narrative that recalibrated my sense of how significant technology develops. It took a long time, with many detours and many players.
The thing is we can't wholly give up our devices and apps, because so many peoples' professional and social lives depend on them. So we have to manage it. It's more like a food addiction than alcohol, since alcohol, you can live without.
Most people are not paying enough attention to this.
Another real danger arises with the so-called "mixing vessel phenomenon," which occurs when different virus strains infect the same host simultaneously, particularly in the context of influenza viruses.
This primarily happens in animals like pigs, which can be infected by both avian and human influenza strains. When multiple virus strains infect the same cells, their genetic material can undergo reassortment - where segments of genetic material from different viruses combine to create new viral variants.
So a corona variant could mix with this fast-spreading avian flu and produce something that impacts humans more.
Viruses are like drunken teenagers -- they behave worse when they mix!