This is a sequel to Daniel-isms: 50 Ideas for Life I Repeatedly Share. Together, these two posts have been my favourite content I’ve ever written — I strongly encourage you to try to create your own list of your frequently shared life wisdom.
This post was awesome! I want to read more posts about people's important life lessons. In ordinary conversation, we don't usually ask the question "what's the most important thing you could tell me that you think could improve my life?"... but we definitely should! I even wrote a post about this phenomenon, along with my own life lessons.
Point 8: Agreed. Also known as 'Tacit Knowledge', and there is definitely no shortcut to this it seems. However, I'd probably preface it with determining beforehand whether or not one necessarily 'needs' to achieve such depths on x topic.
Point 21 you could take even further! This isn't really even an 'opinion', this is a fundamental finding consistently in cognitive science research: free-recall is far (far) harder than recognition, but produces far more effective improvements in learning. This is something, for example, studying with flashcards will reveal very quickly, whereas the student who simply rereads or highlights etc will *feel* and *think* that they are learning, but are not. What you're describing may also be known as "The Feynman Technique"
Re point 22: Aha. This took me back to some poker days. Some professionals will mimic something like this at the table to try and obtain true 'randomness' without bias. I.e., if they feel they should do x 50% of the time, they, for example, would find the nearest clock and would go off whether the second-hand is on the left or right hand side. Smart, as it removes any subconscious bias.
This was a great read. So much so that I had Gemini take a look at it, which you might find interesting and may make for a future blog: https://g.co/gemini/share/7b0e46771c9d
This is such a fun idea! Thank you for sharing it with me. One minor nitpick - I would be curious if the answers change if the model selected was Gemini pro rather than flash.
This paper just published by a team at Apple finds something similar in that reasoning models (LRMs) perform worse at less complex tasks than LLMs, despite being more compute heavy.
It seems they do have the correct solution in their workings out, but can 'overthink' the question and thus end up going down rabbit holes and providing an incorrect final answer.
Re your Q: It would definitely change, though not necessarily for the better and not necessarily as a function of a/the smarter model.
There's a good blog somewhere in my tabs that speaks to this: asking a smarter model a question far below it's intended difficulty level can give less satisfying responses than asking a less smart model*
*However, it would be hard to know if this was down to the output actually being worse or the user just not liking what the output was.
A lot of gold in there, thank you. I wrote many of them down in my quote google doc that I've been keeping for 15+ years. I'll revisit them periodically.
I wish more people felt the urge to record and share life lessons they’ve picked up along the way. You might consider sharing how you compiled these lists. I’d read that!
And the reminder that curation can be as valuable as creation is underated around the hype of startups, which I find valuable and noble, but is not an exclusive goal.
It's on my to do list to read through your back log.
Something very random: in your most recent post 'thank you for recommending me', one of the other blogs is titled 'a long strange trip', in reference to the lyrics of a Grateful Dead song. Before selecting 'not not Talmud' as the name for this Substack, I tried to register 'strangest of places' as the blog name, in reference to another Grateful Dead song lyric (but the person registered with the domain refused to give it up to me). It's a small world with overlapping interests!
This post was awesome! I want to read more posts about people's important life lessons. In ordinary conversation, we don't usually ask the question "what's the most important thing you could tell me that you think could improve my life?"... but we definitely should! I even wrote a post about this phenomenon, along with my own life lessons.
https://open.substack.com/pub/markmcdonaldthoughts/p/anti-elephant-communication
Thank you! Your list is excellent as well. Good values are underrated :)
Here from Reddit. Great read.
Point 8: Agreed. Also known as 'Tacit Knowledge', and there is definitely no shortcut to this it seems. However, I'd probably preface it with determining beforehand whether or not one necessarily 'needs' to achieve such depths on x topic.
Point 21 you could take even further! This isn't really even an 'opinion', this is a fundamental finding consistently in cognitive science research: free-recall is far (far) harder than recognition, but produces far more effective improvements in learning. This is something, for example, studying with flashcards will reveal very quickly, whereas the student who simply rereads or highlights etc will *feel* and *think* that they are learning, but are not. What you're describing may also be known as "The Feynman Technique"
Re point 22: Aha. This took me back to some poker days. Some professionals will mimic something like this at the table to try and obtain true 'randomness' without bias. I.e., if they feel they should do x 50% of the time, they, for example, would find the nearest clock and would go off whether the second-hand is on the left or right hand side. Smart, as it removes any subconscious bias.
This was a great read. So much so that I had Gemini take a look at it, which you might find interesting and may make for a future blog: https://g.co/gemini/share/7b0e46771c9d
This is such a fun idea! Thank you for sharing it with me. One minor nitpick - I would be curious if the answers change if the model selected was Gemini pro rather than flash.
https://machinelearning.apple.com/research/illusion-of-thinking
This paper just published by a team at Apple finds something similar in that reasoning models (LRMs) perform worse at less complex tasks than LLMs, despite being more compute heavy.
It seems they do have the correct solution in their workings out, but can 'overthink' the question and thus end up going down rabbit holes and providing an incorrect final answer.
Would make for a great read.
Re your Q: It would definitely change, though not necessarily for the better and not necessarily as a function of a/the smarter model.
There's a good blog somewhere in my tabs that speaks to this: asking a smarter model a question far below it's intended difficulty level can give less satisfying responses than asking a less smart model*
*However, it would be hard to know if this was down to the output actually being worse or the user just not liking what the output was.
A lot of gold in there, thank you. I wrote many of them down in my quote google doc that I've been keeping for 15+ years. I'll revisit them periodically.
Cheers 💚 🥃
Wow, this is really meaningful to hear. I'm so glad you shared this with me.
Bravo! These are really incredible.
All excellent - best post I've read in a while. Favorites are 10, 29, 34, 38, 45, 47, 50.
Valuable insights. Worthy of attention and reflection.
So inspiring. Everyone should read these once in a while.
I wish more people felt the urge to record and share life lessons they’ve picked up along the way. You might consider sharing how you compiled these lists. I’d read that!
Great, wise post! Great to read pt.2 :)
Number 6 reminded of this by Henrik Klarsson. https://www.henrikkarlsson.xyz/p/unfolding?r=16o30i&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false
And the reminder that curation can be as valuable as creation is underated around the hype of startups, which I find valuable and noble, but is not an exclusive goal.
If you subscribe to mine, you have a good chance of being my 500th subscriber... for real
(Am trying to avoid ssc spaces for the time being, so I won't be checking the subreddit, but thrilled you pmd me to tell me so I could subscribe!]
It's on my to do list to read through your back log.
Something very random: in your most recent post 'thank you for recommending me', one of the other blogs is titled 'a long strange trip', in reference to the lyrics of a Grateful Dead song. Before selecting 'not not Talmud' as the name for this Substack, I tried to register 'strangest of places' as the blog name, in reference to another Grateful Dead song lyric (but the person registered with the domain refused to give it up to me). It's a small world with overlapping interests!