That series of blog posts is incredible, as is all his work. One thing that stuck with me is that while our deep evolutionary past is very important, the majority of humans who have lived have been peasants in an agrarian society
His blog posts are very high quality. It seems however that the average reader ignores his prolific disclaimers about how his work doesn't necessarily generalize and attempting to do so is fraught with peril and attempting to do so any later than the early modern period is laughable.
My personal experience, having had low testosterone for much of my life and then now being on testosterone replacement therapy:
- It’s obvious to me that I (n=1) gained a ton of muscle and lost a ton of fat very quickly without having done dramatic changes in workouts at first
- It also made it waaaay easier to work out. I used to hate it bc I was exhausted all the time, and now I don’t really mind, and sometimes even enjoy it. In actual practice pulling these variables apart is hard
PSA:
T levels have declined markedly over the past generation or two. But when one tests for low T, the goalposts move bc every time a new population is studied for setting benchmarks (generally every decade or two), the definition of clinically “low” is moved to the new 2.5%ile of that study, such that someone in the current 3rd %ile (considered A-okay by most doctors) would have been in maybe 2.4th %ile (considered red alert by most doctors) using the previous benchmarks (made up numbers but roughly right).
(This is a dumb approach, the binary magic of the 2.5%ile (2 standard deviations), and it frustrates me greatly bc i was denied care bc my first test was 2.4%ile and 2nd was 2.6%ile and thus i was told i was fine bc of the 2nd result and offered anti-depressants instead).
Fixing this is not about muscles, it’s about having energy to live life fully and not be cranky.
So if you’re constantly tired, consider testing for this. And use a functional doc, as in my experience from shepherding 5-10 other folks through this process, the standard doc knows little about this stuff and thinks that having a condition like this is binary
Sorry, a bit off-topic, but no change in my life has been more important besides kids, so I try to spread the word where even somewhat appropriate
> no change in my life has been more important besides kids
Quick note to add: if you're reading the above comment and you still want kids, it's important to note that loss of fertility is one of the most significant side effects of testosterone (or any steroid). There are additional medications that can reduce the risk (and possibly reverse it after fertility is already lost), but it's not guaranteed.
Yeah. A lot of the warnings on the side of the box are based on really suspect data from a very long time ago (Abraham Morgentaler at Harvard is my go-to for this) but the point cj raises is legit.
When a man takes exogenous testosterone, their endogenous production (in the testicles) plummets. This is largely fine, but the production in the testicles also leads to sperm production, and that gets lost.
The solution to this is HCG, which stimulates some testicular production. But it is not a perfect solution, and recently CA has made it much harder to use HCG for this purpose (which was always off-label).
The problem with the amyloid hypothesis is most likely not that it is wrong, but that it is incomplete, and I would say that is the problem with this test as well.
I work in neurotech/sleeptech, and AD researchers are using (or want to use) slow-wave enhancement to prevent and possibly manage AD.
However, the test for AD is still a psychological tests along with neuroimaging to look for tau tangles and amyloid plaque build up.
It has been discussed that we may be looking at multiple different diseases which have similar symptoms and without completely understanding the disease itself, we are categorizing them as AD, though they may have different pathways.
Though we can't ignore the challenges to the amyloid hypothesis, we also shouldn't completely throw it out. Most of the experts I've spoken with still believe it is the best hypothesis we have, but that we also should not ignore other possibilities.
It surprises me that some obvious clues to treatment are passed over as I guess they don't fit the politics? Like
>A team of researchers in Jerusalem, he says, decided to look at patients who survived bladder cancer and compare dementia prevalence among patients treated with BCG and those who weren’t. “Do they differ in the rate at which they get Alzheimer’s disease?” The answer is yes – the BCG group appeared to get 75% protection against Alzheimer’s. A number of studies have now found varying levels of protection from BCG, with an average, according to one meta‑analysis, of 45%.
The 'politics' puzzles me. Maybe the head of department got fame for hypothesis A and feels his power or money is threatened by hypothesis B? It's not what science should be about.
(There was an entertaining angry Sabine Hossenfelder
youtube a few minutes ago on the corruption of science just wasting money, but really letting people die of Alzheimer's is worse. https://youtu.be/shFUDPqVmTg)
I think the quote you provided shows exactly why it is so difficult.
When you are comparing a group who survived bladder cancer and then looking at those numbers and figuring out if the bladder cancer treatment had a preventative effect on Alzheimer's, many many things are being conflated together.
You initially have survivorship bias of those who were alive after bladder cancer. How many of the bladder cancer people had AD, or another disease? Though I'm sure the BCG treatment helped in their treatment of bladder cancer, did these people respond better to the bladder cancer treatment because of other factors? Did they have better lifestyle than those who did not, less inflammation, better diet, etc etc.
How many women were in the study? AD affects twice as many women as men. If AD is an infectious disease, why would this be? Type 3 Diabetes is a strong candidate as an alternative theory to AD, but Type 2 Diabetes, though more damaging to women, does not occur at twice the rate as in men? So why would this be?
Again, I'm not saying other theories are wrong, but everything is VERY difficult to prove.
The Amyloid hypothesis, which I would also label as the sleep theory of AD, is tied to reduced glymphatic function, reducing the brain's ability to clear metabolic waste. Why would this affect women much more prominently than men? Motherhood and Menopause, both of which are very disruptive to women's sleep.
Of course maternal diabetes can also play a significant role.
I completely agree with Sabine Hossenfelder and you that improvements need to be made to the scientific process, including publishing of negative findings, and improved open discussion regarding published papers. I suggested to someone recently that they create a HackerNews of research papers and we don't just have citations, but open discussion of people expert in the space.
There are letters in publications, but that is not directly tied and linked back to the paper. A friend was developing something similar to this years ago, but was never able to find enough traction or a business model that makes it work.
I think that was an early and somewhat naive understanding of why. I wouldn't say it is not true, but the 5 year lifespan difference resulting in 50% more cases of Alzheimer's in women seems an oversimplification.
There‘s also a weird relation with the oral microbiome. Probably not that surprising, seeing how much translocation is actually happening between our microbiomes.
Just an n=1 ofc, but someone in my family got all their remaining teeth pulled and replaced with implants, and afterwards, within 3 years went from asymptomatic to dead from AD. The progression was mind boggling. I‘ve wondered for a long time whether the oral work was related to this - the surgery caused so much damage that it certainly exceeded capacity to heal, opening the doors wide for any pathogens.
I agree to a point, but you also have to consider, why did they have to get their "remaining teeth pulled"? It is entirely likely the AD was present before the teeth were removed.
I had an uncle with AD, and he hid it VERY well for quite a few years. We would have thought he was asymptomatic, but we suspect it was just that we didn't know.
I guess it only takes one bad pathogen to get in and it's hard to know what happened without scientific testing.
The Guardian article I linked in another comment has for one guy "sample of his cerebrospinal fluid was taken and revealed a fungal infection..." which it shows it's possible at least to some extent to test for this kind of stuff. (article https://www.theguardian.com/lifeandstyle/2024/dec/01/the-bra...)
It seems a bit of an under explored area. I suspected something like that with my dad who had a long deterioration of nerves/mind but current medics don't seem to regard it as a thing to do. It's more get your affairs together before the care home/death. While the guy in the Guardian article guesses that likely more than 50% of dementias are treatable infections.
"The problem with the amyloid hypothesis is most likely not that it is wrong, but that it is incomplete,"
If the hypothesis is that amalyoid causes AD, then I think we've disproved that (even with your statement that it's incomplete). If I remember correctly, there are individuals who have amalyoid without developing AD, or who have had their amalyoid levels reduced without improvement. At this point, it seems that amalyoid is more of a symptom than a cause. But you are correct that the data is dirty - many studies have not tested for amalyoid itself, instead relying on clinical diagnosis, and subsequent studies are finding at 25% of mild to moderate AD may be other forms of neural degeneration. So many of the AD studies out there focusing on amalyoid reduction are garbage because many of them happened before being able to use imaging to test amalyoid levels.
This is what I mean by "we are likely looking at different diseases". Something that would be diagnosed as AD, but without amyloid levels (if it is decided that this the correct marker to look at).
It's as if we were to look at diabetes, and then not recognize there are (at least) two types of diabetes, with VERY different causes, and recommended treatments.
Damn! I know nothing about AD and while reading these comments you mentioned sleep tech. I didn’t understand the connection or what amyloid hypothesis even was, so looked it up.
As someone who’s on < 4 hours of sleep a day with at least one day for 36 hours straight awake a week, umm… should I be worried?
Yes, you should be very worried. Imagine accumulating trash in your house—you take some out, but you never throw away all of it. Each time, some remains. Now, imagine what happens after a few decades. In this example, the house represents your brain.
Agree strongly with you. This statement make Alzheimer’s into a wonderfully “simple” monogenic disease like Huntington’s but all age-related disease have many complex interwoven weaker and stronger causes—-even Huntington’s disease in which the same mutation type (numbers of CAG mutations in neurons) can cause symptoms over a 20 year range.
Too bad that headlines are inherently short and sometimes misleading. Simple sells.
Is it not still the case that they do correlate, and therefore the article talking about them as biomarkers is not making the mistake you think it is (as if it was talking about them as the thing to get rid of to prevent Alzheimer's)? Because "lead to" is not the same as "causes".
Or has latest research shown that even a non-causal link should be dismissed?
Ok “lead to” might not be exactly equivalent to “causes”, but it’s not even accurate because tau doesn’t ALWAYS lead to Alzheimer’s. The correlation is not strong. People have tau and no Alzheimer’s and other people have Alzheimer’s symptoms and no tau. So what good does it do you to test for tau? It’s a higher probability of Alzheimer’s, but not nearly definitive. I suspect that if this test is productized, this nuance will be lost on many doctors.
Seems like a good test to give someone so you can scare them into taking a drug that also has very little evidence of effectiveness.
It can be used that way, but not always. In fact, traditionally your way is the less common usage.
A path of breadcrumbs leads to the gingerbread house, despite the house not existing because of the path (or a road leading to a house, to get away from witches).
Unless you mean that in medical research it's universally agreed to only use it in that way? I'm sceptical that's the case but could be convinced...
(Either way, "lead to" obviously isn't at all clear enough if they did intend to mean "points towards, without causation".)
I've seen this point brought up many times over the years and yet research in this direction seems to continue, so I think there must be more to it. I asked chatgpt about it, and it claims that the controversial paper was influential but it was more about a certain sub-hypothesis that got called into question after that rather than the entire thing.
Now I know that bringing up chatgpt is frowned upon here but I thought I should make an exception for this case as its not so easy for me to answer otherwise.
You might enjoy “How Not To Study A Disease: The Story of Alzheimer’s” by Karl Herrup (2023, MIT Press). Great sad overview of the hegemony of the Abeta hypothesis that now has a Tau hypothesis buddy.
I went to a Jesuit high school but am not religious now. But I still have a soft spot for the Jesuits. They tend to be liberal (relatively speaking) and taught extensively about social justice, and generally are very education-oriented.
I’d say it’s pro economic development. Like they express concerns around the decline of anti-trust enforcement.
I’m sure it’s true that they used to advocate dictators, but in the 30 years of reading it as my primary news source, they’ve always seemed to me to be very consistently on the side of liberalism (in the older sense of the word) and very concerned about democracy
I know very little about lawyering, but I could imagine a UW-alum or Seattle-area lawyer advising pro bono bc of generosity or good publicity on a very newsworthy case
Anyone on here friends with a UW-alum or Seattle-area lawyer who might be interested but doesn’t read HN?
It's very common in the UK. The most visible part of the unions is running social activities, often bars and events, but they can also provide legal advice to their members.
Not undergraduates. The "student union" in this context is 100% a social entity.
Graduate students sometimes are part of unions, but usually only if they're also employed by the university (somebody paying full tuition for an MBA probably isn't in a union, but a doctoral student teaching or doing research might be).
Undergrads doing part-time work at the university to pay bills (dining hall, bookstore, etc) could be, in theory, but probably aren't.
Graduate students (who teach, do research, and some administrative tasks for the university) occasionally do. What American schools usually have is a “student government” which is approximately 90% roleplaying as elected officials, and the remaining 10% is deciding which banquets they should host themselves to spend the small budget the university gives them.
Even if this student isn't a member, the local graduate student union (https://www.uaw4121.org/) would probably be a useful ally. All TAs are in the bargaining unit, and UW CSE has _a lot_ of undergrad TAs, so I wouldn't even be surprised if this student is a present (or former) member.
Possibly this is not the actual dress code? Or I'm missing something.
3.a. The following is acceptable for men players, captains, head of delegation.
-- Suits, ties, dressy pants, trousers, jeans...
3.b. The following is NOT acceptable for men players, captains, head of delegation.
-- Beach-wear slips, profanity and nude or semi-nude pictures printed on shirts, torn pants or jeans...
Here is a short example that came up for me last week.
I had a set of documents I wanted to classify according a taxonomy that is well known (so it is exists in the training data of all the major llm models I tested)
If I have prompt like, `You are an expert classification system. Using the Classification Approach Foo, consider the following and output the category in JSON format, such as {"class":"bar"} `
This works ok, but it works much better if I tell it to output {"class":"bar", "reason": "baz"} and improved with some other approaches like adding "related_class" or "parent_category" which would otherwise be redundant.
Also including some few-shot examples helped, but the biggest benefit came from the "reason" field. Trying justification or other synonyms seems to produce the same output.
Have you tested moving the "reason" field before the "class" field? That may encourage better CoT instead of having the model justify the class after it already picked it. Anecdotally, I saw a 5% boost in performance from a NER system from having the model output the entity's class at the end rather than the beginning.
Speaking only for myself these ideas are a combination of things I've seen scanning new papers and informal discussions with other people working in the area. Feel free to shoot me an e-mail though, maybe I can point you somewhere more specific.
Edit: The "verbosity sink" name is inspired by the idea from the paper below although they're not actually at all the same thing.
I was in the S12 batch. A couple years later we were demonstrably failing. I emailed the general YC advice line or something similar with a question, and pg responded with a thoughtful comment that was genuinely useful
I don’t think there’s any chance he saw any economic value in us, he took that time because he cared about startups and well-meaning failures like us
It’s a small thing, yes, but I’ve always thought it said a lot about him
Agreed. In a similar vein, when our YC company was failing I had a great call with Jessica Livingston, and looking back it was purely moral support and not anything economically motivated.
