Hacker News Reader

>he poorer a person is the larger a percentage they are willing to risk

I don't agree with it. Poor people tend to be very risk-averse (they teach their kids to get "stable jobs" above everything else), and this puts them in a self-reciprocating loop of poverty. Many rich people who i know became so by repeatedly getting involved into "bold" things looked totally bullshit, and usually indeed was - apart from 1-2 in their lives that wasn't, and made them millions.

I think you're a bit burnt out from physics, thus this rather cynical view of physicists' approach to an important economic problem.

The mathematical models you find in the works of many highly regarded economists are even more simplistic, more vulnerable to criticism.

"When you have such hubris and make such simple mistakes..." that pronoun "you" should be the economists!

That economists make mistakes, and that even well-regarded economists espouse simplistic models, is not evidence that physicists might be any better at it.

Economic models are simplified in such a way to describe causality and relationships between variables but aren't actually expected to work 1:1 with real world numbers.

That's why econometrics is a thing, that's why most economists nowadays are taught statistics, computer programming, advanced maths, etc... It's why there's a different process used to describe why things happen in a paper vs. predicting real world numbers.

> the poorer a person is the larger a percentage they are willing to risk

As a percentage, sure, but in real terms the poorer person doesn't have 'much' to lose (to be regained) over someone extremely rich risking the same percentage. The potential rewards are also likely to be immensely larger in percentage for the poor person too.

But this is also relative risk against wealth in context. Comparing poor against poor there will still be comparable poor people with completely different risk profiles (basic example is one with children vs one without).

The model I imagine must primarily establish reasonable/full context before determining any value factors.

As a percentage, sure, but in real terms the poorer person doesn't have 'much' to lose (to be regained) over someone extremely rich risking the same percentage.

This is questionable. The assumption that a poor person can more easily regain their stake is shaky---a poor person who loses their money may become homeless, which could in turn make it harder for them to be employed, or if they are a trades person maybe they lose their tools or can no longer afford to rent them. By contrast, a rich person has access to credit lines and investors in their social network. For the poor who aren't wiped out below a critical threshold, then, it may be true, but you're not taking into account the significant number who drop below that threshold and may never recover.

I would rephrase the original statement, "the poorer a person the larger percentage they are willing to risk" as "the poorer the person, the larger percentage they are forced to risk if they want to achieve comparable returns on a percentage basis."

Rich people do not have access to credit lines if they don’t have assets to back the loans. And their rich friends certainly aren’t going to invest in someone who just lost a ton of money.

That's better than a poor person, who does not have access to reasonably priced credit even in the unlikely event they have the collateral. And in reality, the rich often do have access to unsecured credit. They have that credit because they've made money in the past, and even if their current fortunes have turned, creditors and investors will often continue to invest in them.

One of may favorite quotes about models is (paraphrased): All models are wrong, but some models are useful.

I see it most often attributed to George Box, a statistician, but it's not completely clear he was the originator. In physics you have a luxury of verifying models at many empiric touchpoints via data collected in controlled or precisely measurable conditions. Comparatively in economics it's difficult to establish you have a useful model because the empiric data is difficult, and with unmeasurable aspects around human choices.

It's analogous to weight-loss arguments.

It is true that calories = calories-in - calories-out.

It cannot not be true.

It just isn't helpful.

I found it helpful. In a world where a million distracting influences are promising cheats, shortcuts, and easy ways to get to the goal, a single true statement to focus attention and wipe away the bullshit is incredibly useful.

Sugar and insulin response are critical to weight loss. The more overweight, the more you're insulin resistant, the harder it is to actually lose weight. Once you hit metabolic syndrome you've got a big problem on your hands. The CI-CO model is not sufficient to model weight loss effectively. Especially since with caloric restriction your BMR slows down to match your reduced intake (which doesn't happen with fasting, FWIW).

tl;dr: The only time CI-CO is truly accurate is when CI=0.

That's an extreme overgeneralization that's wrong in just the right places to be misleading. While there is a reduction in energy expenditure associated with the reduction of caloric intake, it is not equivalent to the reduction in caloric intake.

For instance, in a study of obese women, a reduction of (4.3 +-(4.6) kcal/kg) was observed in a calorie restriction diet of 20kcal/kg

Similarly, in a different study, a 30% calorie intake reduction was linked to only about a 5% reduction in total energy expenditure.

CI-CO is not a perfect model, in that it doesn't perfectly estimate weight loss rates, but the general concept holds true.

https://academic.oup.com/ajcn/article/85/1/73/4649429

> That's an extreme overgeneralization that's wrong in just the right places to be misleading. While there is a reduction in energy expenditure associated with the reduction of caloric intake, it is not equivalent to the reduction in caloric intake.

I didn't mean to imply that it was 1:1, just that the model doesn't hold entirely.

> CI-CO is not a perfect model, in that it doesn't perfectly estimate weight loss rates, but the general concept holds true.

In general, I suppose, however most people aren't trying to lose weight but rather fat. Consuming sugar causes a spike in insulin, and insulin triggers the uptake of sugar in the blood by adipose tissue as fat, and actively prevents the release of fats stored therein. [1]

So in the most literal sense, CI-CO works, in that (net of the 16% reduction in TEE relative to your reduction in CI per your data) you're going to need to get the calories from somewhere. However, if you eat 30% less calories than your BMR entirely as sugar, you're much more likely to be burning alcohol, glycogen and protein than fat, certainly in the near term.

If you combine that with the "eat 6 meals per day" fad and you're in for a seriously tough time losing fat. On the other hand if you at 30% less calories than your BMR entirely as sugar in one sitting over 20 minutes and fasted the remaining 23h30 you may actually lose fat, since that'll give your body time with low insulin levels.

Let me re-state: CI-CO is a very inadequate model for losing fat.

[1] https://www.diabetes.co.uk/news/2018/jan/drop-in-both-insuli...

CI doesn't have to be 0. That's total nonsense. You think if you eat one grape per day suddenly you won't lose weight, or something? Who said anything about 0?

CI-CO is accurate if CI<CO. There is absolutely no way it cannot be. This doesn't depend on any kind of reasoning about biology or insulin or hormones at all. It's the law of conservation of energy.

That's the beauty of it. It always works. If CO<CI, you will lose weight. There has never been an exception to this in the history of life, and there never will be.

Even fasting, which you mentioned, only works as a means to an end. And that end is CI<CO. Research shows that the actual timing of fasts is irrelevant; the same calories eaten in 6 hours per day have the same effect if eaten over 16 hours of the day. But when people fast they tend to eat less overall; it's all that matters.

So many people try to follow your path, and figure out a way to manipulate complex metabolic systems they don't understand to try to wheedle the body into dropping pounds. It works only rarity, and 98% of people on this path are just deluding themselves to avoid facing the fact that they need to get CI down under CO. So they take refuge in complexity, endlessly finding ways to rationalize away the need to do the actual work.

That phrase is usually used to remind people that eating less is as if not more important than exercising to lose weight. It takes a lot of effort to burn a Calorie and next to no effort to consume one.

Aka “you can’t outrun the fork.”

A common joke on weightlifting and bodybuilding forums when someone asks how to get shredded abs to to tell them to add 'fork-putdowns x F' to their routine. x F is the format used to mean "repeat until failure", and it's funny because it really is the best way to get all of the fat off your midsection so your abs are visible.

How are you defining a "calorie"? The number printed on the package? Because that number is basically meaningless.

That number happens to be remarkably accurate. More accurate than required by law in fact. It also happens to accurately predict changes in a person's weight if you know everything they're eating.

A question I've had that I never really looked into or found an answer to:

How does calories in vs calories out work regarding the energy left in food that is not extracted (i.e. the caloric value of poop basically).

Is digestion / energy extraction pretty consistent throughout the population or do the specifics of people's digestive systems play a role in how many calories are actually available to the body?

Basically, would a constipated person's digestive tract pull more energy out of the food since it is in the intestines longer?

While it certainly differs from country to country, FDA allows a discrepancy of 20%. That's significant. There have been studies that showed the labels to be off by 25%.

And even if the numbers were accurate, they're still rule-of-thumbish, because human metabolism doesn't simply ingest all energy available in a food. For instance you'd absorb roughly 40% more calories from peanut butter than from actual peanuts (of the same calorical value).

There are more reasons why counting calories isn't nearly as accurate a method as people tend to assume, but these two are really sufficient.

Isn't the low carb argument that for proteins more is passed and don't need to be burnt off, but for carbs almost all calories must be burnt off? How many calories do I take in and therefore must burn off thru exercise if I swallow a penny?

I think the low carb argument is that processed carbs mess with your blood sugar levels, making you hungrier, but the discourse around nutrition is messed up enough that there are many low carb arguments, some of them mutually contradictory. As for your penny, zero, unless I'm greatly mistaken, which is why pennies don't come with nutritional labels.

THE low carb argument is ketosis. That's not even controversial. If you look it up on Wikipedia, it's described in a very aseptic way, it just a basic fact of metabolism. If you look at low carbs diets articles instead, a cloud of disinformation appears.

Thinking in isolation and with pride ends in being an idiot.

- GK Chesterton

Albert Einstein did much of his greatest work in isolation, and he was quite proud, too.

If you read Isaacson's biography, it's hard to come away without feeling like Einstein's insights were a product of his time, that he was in some sense no more a vessel than each of us, and that vessel was continuously exchanging contents with others: his father, the family's weekly dinner guest (a local student), the classmates he cribbed notes from, the professors who found him frustrating, the playboy he tutored (and learned to smoke cigars from), his wife, the patent applications on time-keeping devices he read as a Swiss patent clerk (and his train-time-keeping metaphor), his many collaborators (Lorenz perhaps most of all), the early conferences in Paris, etc.

I am no Einstein, to be sure, but I do particularly recognize the experience he had as a teen, learning the math and engineering of his father from his father. His father and uncle, it's worth noting, built electric power generators. Mentally building an effective model of fields was surely something Einstein had to learn early, and perhaps without the math first.

And, yet, in 200 years he may well be regarded as an idiot.

Why? Newton is still regarded as one of the greatest scientists ever.

Note that he too did most of his work in isolation, and he too was extremely proud.

Exceptions only prove the rule.

My first exposure to really opportunity to contemplate economics, in even a qualitative sense, came with Dierdre McClosky's essay in the early days of "post-autistic" economics (1-3). Her observations about the flimsiness of economic modeling, verging on simply serving as a veil naked capitalism, struck me as the sort of informed cynicism I could hold onto. 17 years later, it's her disdain for economic models in comparison to physical models that sticks with me.

(1-3): I honestly forget which of these it was, but the key term I search for is "mind-breakingly"

https://www.deirdremccloskey.com/docs/paradigm.pdf

http://www.deirdremccloskey.com/docs/pdf/Article_320.pdf

https://www.deirdremccloskey.com/articles/stats/stats.php

There is a crucial difference between physics and economic theories. Physics is a hard science. Physics works.

> the poorer a person is the larger a percentage they are willing to risk

Utility function is logaritmic.

Is there any evidence that it is?

Pretty common assumption. [1]

The need for money saturates slowly. Think about winning 1000 dollars (nice dinner pehaps), 10 000 (better car), 100 000 (larger appartment), .... , 10 000 000.

At some point it doesn't matter if its 200 000 000 or 300 000 000.

On the other hand a lotter ticket is 1$. What is the utility of 1$ - a can of coke pehaps. But the perspective of winning $1 000 000 - thats a nice idea to hold for a few days... So many people do despite the slim chances.

[] https://economics.stackexchange.com/questions/23952/is-the-u...

[] https://www.forbes.com/sites/rogerkay/2013/08/05/how-the-mar...

That's evidence for a concave utility function wrt income, not logarithmic per se.

But your first link does cite evidence that _if_ utility is CRRA, we are not far from the case where utility is logarithmic