commit 6d0ad1e02992c8ba4432547a7acbd2c5f65030aa
parent 3b5ead96ea94ef5ac56fb25e0fbb6e82527506c5
Author: Andrew Laack <andrew@laack.co>
Date: Fri, 28 Aug 2026 22:55:51 -0500
Some wip posts
Diffstat:
3 files changed, 139 insertions(+), 0 deletions(-)
diff --git a/posts/wip/how-i-learn.md b/posts/wip/how-i-learn.md
@@ -0,0 +1,12 @@
+# How to Learn
+
+A few things are agreed upon in terms of their efficacy in aiding learning:
+
+1. Spaced repetition
+ - Do things over time that exercise memories
+2. Retrieval practice
+ - Pull from memory to answer questions, write about a topic, remember what you read / saw
+3. Interleaving
+ - Helps with dis-ambiguating similar topics
+4. Self-explanation / interrogation
+ - Ask questions to yourself, explain topics to yourself (consider another person and how you might explain it to them)
diff --git a/posts/wip/should-i-be-using-llms-for-learning.md b/posts/wip/should-i-be-using-llms-for-learning.md
@@ -0,0 +1,16 @@
+# Should I Be Using LLMs For Learning?
+
+I originally considered writing about whether I thought LLMs would 'revolutionize' education, but I was recently reminded how difficult it is to predict the future. An evaluation of the merits of LLMs for education does not lead to a conclusion about the 'revolutionization' of education, and given the investment into AI, along with uncertainty about payoffs, it's even more difficult to predict where such things will invariably lead to.
+
+Instead, let's tackles a specific issue, "should you be using LLMs for learning"? I'd like to start this with my personal view on the matter: I would not recommend using LLMs for learning. Without considering the merits of LLMs for learning, there are ethical reasons I would warn against using them for education. Specifically:
+
+- Support of proprietary software ecosystems / networked software
+ - Most language models worth learning from would likely be prohibitively expensive to run so one'd use someone else's compute which is functionally proprietary
+- Intellectual property theft
+ - I am not opposed to such things, but many are, so perhaps it's worth considering
+- Fear-mongering
+ - The fear mongering being done by these AI companies about job loss and the ideological push that you'll be 'left behind' are doing societal harm
+
+# Should I Pay a Tutor
+
+Also probably not. Seems like they aren't better than step-wise learning https://www.tandfonline.com/doi/abs/10.1080/00461520.2011.611369.
diff --git a/posts/wip/tracking-lifts.md b/posts/wip/tracking-lifts.md
@@ -0,0 +1,111 @@
+# Tracking Lifts (How I Broke a Years Long Plateau and Became an Intermediate Lifter)
+
+## Levels of Lifter
+
+A few definitions by layer
+
+### Time-based
+
+- Beginner
+ - 0 - 2 years
+- Intermediate
+ - 3 - 6 years
+- Advanced
+ - 7+ years
+
+These aren't 100% agreed upon, and some may argue advanced might be 3+ years. Simply going to the gym doesn't mean someone understands how to lift weights, even if they've gone for a long time. On average, you can expect people who have been lifting for longer to be more experienced, but this isn't perfect.
+
+### Metric-based
+
+The big three are sometimes used to group categories of lifters based on % of bodyweight their 1RM is for the given lift. Below are rough guidelines for bench press:
+
+- Beginner
+ - 0% - 100% of bodyweight
+- Intermediate
+ - 100% - 150% of bodyweight
+- Advanced
+ - 150+% of bodyweight
+
+These can be argued, and to make this more accurate you can group individuals by age range, but even this is limited. There are many roided out individuals in my age group who don't really understand lifting, are blasting anabolic steroids, and can easily outlift me on the bench, but I'd argue these individuals are not as advanced as I am in the gym.
+
+
+### Experience-based
+
+My preferred definitions for beginner, intermediate, and advanced are as follows:
+
+- Beginner
+ - Evident improvements month to month without the requirement of programming changes
+- Intermediate
+ - Broken through their first lifting plateau via programming changes
+- Advanced
+ - Everything has to be dialed in to gain muscle
+
+## My Story
+
+My age is between 22 and 27. I started lifting weights when I was 12 when my father gave me a rudimentary PPL-esque workout plan. Inspired by Youtube's weightlifting scene (I specifically remember the hodge twins, connor murphy, and rich piana from that time period) I stuck to that
+
+I stuck to this plan for a few months before stopping. I don't remember the specifics of this, but it's likely my consistent schedule got interrupted and I simply didn't get back to it. After a nudge from my father to get back into it a few months later, I started lifting weights again. From that point on, to the best of my knowledge, the largest number of consecutive days I didn't do strength training was 2.
+
+A few years ago I started adding exercises to the core set my father prescribed as part of my plan, but prior to that I had religiously stuck to the plan.
+
+A month ago I realized I'd spent the past few years only improving marginally. I couldn't quantify the amount I was improving, but it was very minimal. I realized this when I went for a 225 1RM again after first hitting 6 months prior and barely eeking it out again. I suspect most of the gains I got in the past few years was going from a home gym to a commercial gym which also forced me to make my sessions more concise. This resulted in a cut in rest times between sets along with an increase in volume. This likely broke a plateau, but I didn't realize this at the time, and the plateau was just pushed up slightly higher.
+
+Since I didn't consciously identify the plateau at the time, I can't say this was me breaking into the intermediate lifter category. I did identify the plateau when doing the second 1RM of 225 though. The reason I was able to identify this was because I remembered a prior weight I did. After this, I decided to start tracking all of my lifts.
+
+Over the past few months, this has led to sizeable lifting gains, and an increase in my effort to match or beat prior comparable sets.
+
+## Enhanced Self is the Enemy of a Good Life
+
+Despite these improvements from tracking, tracking my BP has decreased my quality of life. As such, I stopped doing this after a few weeks. Tracking my blood pressure only served to stress me out for no good reason because my BP was always healthy except when I would get nervous about my BP reading being elevated, which would then happen, but this was never dictated by diet, and always dictated by thought. (draw parallel to sleep).
+
+To this end, I think it's only a good idea to track things that 1) help you improve your performance 2) actually matter.
+
+- tracking porn
+ - https://www.bradstulberg.org/p/the-enhanced-self-is-the-enemy-of
+
+---
+
+- what to track?
+ - things you can change
+ - I can change my weight / bp, but like really?
+ - probably not a good idea
+ - track things latently?
+ - only after things will be noticed should I track them?
+ - what about only things that are incremental?
+ - perhaps this?
+ - like if you are moving towards a pr then maybe track that
+ - track the most important stuff?
+ - like no one cares about slowest mile probably, maybe just like start to end?
+
+> Until recently, you received health metrics once a year at your annual physical. You understood your social circle based on the number of people who showed up at weddings, birthdays, and holiday parties. Professional success was measured on yearly lists. Athletes knew where they ranked based on a few events they peaked for every season. We still measured and compared, but there were long stretches of time between those measurements and comparisons. And in those long stretches of time, we could forget about ourselves and how we stacked up. That is no longer the case.
+
+This made me think we should only track things on longer timescales or perhaps only non-digitally. I think the ladder is likely incorrect, and merely a facilitator of the former. that said, I do think there's some merit to the former. Perhaps its short term tracking of things that are weakly correlated? Like it seems fine to track lifting because in the short term it's highly correlated. But something like blood pressure is much more stochastic, as is sleep. You don't really get very very good at sleep and then get huge amounts of it over time, it's a test each time. If it's a test each time, it shouldn't be tracked that frequently. This holds for the bp and other tests as well.
+
+This also allows longer term tracking of bp and body weight which will miss the arbitrary fluctutations. When lifting there are very small fluctuations over time so it seems permissible to track. This might also be that bullshit rationalist thing where you decide on something and then try to find a reason for you to do it. I want to make shit tons of money -> this is good for the world because I can give it away so I would be stupid not to.
+
+Or perhaps it's this: If there are any consequences to tracking, like taking some non-trivial amount of time, increasing stress, perhaps degrading biomarkers, do it sparsely.
+
+> orthosomnia
+
+Worsening of sleep due to tracking
+
+---
+
+General rule: If the cost of tracking something is greater than the benefits, don't do it.
+
+If it stresses you out unduly, don't track it. If it takes a lot of time, don't track it. If it degrades the biomarker, don't track it.
+
+---
+
+
+> Maslow used the term peak experience to capture transcendent moments of joy, awe, and deep fulfillment
+
+> The inverse is also true: studies show that an excessive focus on the self is associated with choking, be it on the playing field, in the boardroom, or in the bedroom. Look more broadly across all of life, and research shows excessive self-focus is linked to unhappiness, anxiety, and depression.
+
+---
+
+Prescription: Write it down and don't give a shit. When will you give a shit? Some amount of time later. A month, a week, a year, whatever, but it isn't now. At that point look back and see how things have gone.
+
+High Level: Check them less frequently.
+
+Perhaps only track lifts once every other week? This would be some sort of mesocycle tracking.
