HNHacker News
TopNewBestAskShowJobs

czl

240 karma · joined January 14, 2016

submissionscomments
czl··on Are animals conscious? New research
>What is most characteristic of intelligence is intentionality.

Intentionality refers to the quality of being directed toward an objective or purpose. It involves conscious decision-making and goal-directed behavior. Given that insect have intentionally do you consider insects intelligent?

While intentionality might motivate and guide our actions, the ability to foresee the results of these actions is what defines intelligent decision-making. Prediction requires a deep understanding of cause and effect, which is a hallmark of intelligence. When we predict, we infer potential outcomes based on our knowledge and reasoning. This process involves not just abstract thinking but also practical application. Accurate prediction involves the synthesis of various cognitive functions, including inference, analysis, and reasoning.

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> But wouldn't the logic you employ apply to any strategy?

You ask a good question.

Consider a game like rock-paper-scissors. The goal is to predict your opponent's move. Among all possible strategies, the random strategy is unique because it can't be predicted. Any other strategy can be anticipated by a smarter opponent. Being random means you can't lose to a smarter opponent.

In trading, buying low and selling high requires prediction. Any strategy that relies on prediction becomes predictable to smarter opponents. Buy-and-hold index investing is the only strategy immune to this exploitation because you're effectively betting on everything. With passive index investing, your investments grow at the rate of business growth, making this strategy special.

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
HN had an error during saving and when I attempted again I picked the wrong message and can not change or delete it now. I will add an extra answer later.
czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> I am just stunned by your worldview which I find is wrong in so many ways.

100% my reaction towards your comments as well.

> It's no wonder that you couldn't muster the intelligence to come up with any profitable strategy at all.

Studies have shown that over the long term, the majority of actively managed funds fail to consistently outperform their benchmark indexes after accounting for fees. With facts like these my intelligence says stick to index investing.

> Moreover, you go about confidently asserting and spreading your disinformation, attempting to brainwash others with your false beliefs and lies.

I am sharing commonly accepted knowledge. It is commonly accepted that buy low sell high trading strategies are like perpetual motion machines. This does not stop some poorly educated honest inventors from inventing them however. Con artists also “invent” them and offer them for sale or offer their expertise to help build one for yourself. Which of these you are I am not sure yet. Early on it can be hard to tell since a good con artist is subtle.

> Like I said, 25 is between 10 and 30. It depends what exactly the strategy's performance is, and we don't know what it is for various reasons. I look at it day to day.

You realize investments are not judged by returns but metrics like returns per risk such as sharpen ratio, sortino ratio, etc? If risk did not matter leverage can trivially multiply returns. TQQQ for example can look attractive till you realize a daily ~30% index drop wipes you out as happened with oil versions of such funds during COVID pandemic.

> You with your limited worldview are presupposing that all strategies work because others haven't found them.

Once a strategy is known, it gets rapidly arbitraged away as traders compete to exploit the inefficiency. Widespread adoption of a strategy changes market dynamics in ways that invalidate the original premise. Strategies that become crowded trades are vulnerable to severe losses if sentiment turns and everyone rushes to exit at once.

> That's not how they all work. Some work despite the fact or even because of the fact.

Yes! Pump and dump schemes and similar fraudulent trading strategies exploit popularity to manipulate markets. These scams often involve self-proclaimed "gurus" who heavily promote a stock or cryptocurrency, artificially inflating its price. They lure unsuspecting and inexperienced investors with promises of easy profits, diverting them from proven investment strategies like index investing.

The more popular and influential the guru, the greater their ability to temporarily move markets with their recommendations. In essence, the guru's predictions can become a self-fulfilling prophecy in the short term, as a herd of enthusiastic followers rush to buy the promoted asset. However, this popularity is manufactured and fleeting.

What the followers don't realize is that the guru has already quietly accumulated a position beforehand at a much lower price. As the guru's devotees drive up the price by buying en masse, the guru sells their holdings at the artificially inflated prices, pocketing significant profits. The asset's price then collapses, leaving the guru's followers with substantial losses.

In this way, pump and dump schemes enable unscrupulous individuals to exploit the trust and capital of unsophisticated investors. Regulators are increasingly cracking down on these scams, but they remain a persistent problem, particularly in lightly regulated markets like cryptocurrencies. Investors should be highly skeptical of "get rich quick" trading tips, especially those popularized by social media influencers or self-declared experts promising unrealistic returns.

> Wrong again. Professional traders specifically discard such retail noise, although I suppose it could be used in certain contexts.

So why do some professional traders pay to trade with retail order flow? They pay for it so They can “specifically discard such retail noise”?

Obviously not. Professional traders try to exploit retail order flow by "front-running" - getting in front of large retail orders that are likely to move the price. If they can buy right before a large influx of retail buy orders hits the market, they can then sell into that buying pressure for a quick profit. Order flow can also give a view into where retail traders are placing stop-loss and limit orders. Professional traders can use this to their advantage, triggering stops or fading popular limit order levels.

> The number has no basis in any zero sum game; it is whatever the buyer last paid.

Stock trading can be considered a zero-sum as one trader's gain is another trader's loss. When someone buys a stock and the price goes up, they profit, but the person who sold it to them missed out on those gains. Conversely, if the stock price goes down, the buyer loses money while the seller avoided those losses. On any gi rn day of trading the total wealth of all participants remains the same, with money just changing hands between them.

In the long run however stock trading is actually a negative-sum for traders.

Traders pay fees to brokers, exchanges, and other intermediaries for executing their trades. These costs eat into the overall profits, making the total wealth of all participants decrease over time.

There is typically a small difference between the highest price a buyer is willing to pay (bid) and the lowest price a seller is willing to accept (ask). This spread represents a loss for the overall market, as it's a cost that doesn't contribute to any participant's profits.

Time and resources devoted to trading could have been used for other productive activities, such as starting a business or investing in education, which could have generated wealth for the overall economy.

> If your lies were the truth, bitcoin wouldn't exist, and gold wouldn't carry value either.

Tell me why are prices rarely advertised using gold or Bitcoin? Might it be because these are speculative assets rather than currencies. If they were widely accepted as currencies, you would see prices consistently posted using units of gold or Bitcoin. However, this is not the case due to their highly volatile value, which is a result of the lack of active supply management.

The primary purpose of a currency is to maintain stable prices while discouraging the hoarding of the currency itself. This is achieved by controlling the money supply to maintain a small, consistent rate of inflation.

When excessive debt accumulates in the economy, inflation may be strategically increased to help clear that debt.

This is a fundamental aspect of how modern monetary systems function, and it is well-understood by those with an education in economics. There are no surprises here.

It is generally unwise to store wealth using a fiat currency for extended periods, as it is not designed for that purpose. Instead, the main objective of a fiat currency is to facilitate transactions and promote overall economic health. Other investment vehicles, such as real estate, stocks, or bonds, are more suitable for long-term wealth preservation and growth.

> Gold was used as a currency for thousands of years, also by the Romans and by the US with a gold-backed dollar.

Societies used to not care much about washing and keeping clean for thousands of years. For this reason we should go back to that? The fallacy in this argument is an appeal to tradition or appeal to antiquity. Just because something was done a certain way for a long time, does not necessarily mean it is the best or most appropriate way to do things now.

> Anyway, as a user of money, I couldn't care less about the destructive agenda of economic growth.

Economic growth today is essential for creating a better future for coming generations. By expanding productive capacities, raising living standards, and spurring innovation, we lay the foundation for a world of greater prosperity and human flourishing.

Growth enables critical investments in education, research, and governance that pay long-term dividends. We have a moral obligation to be good ancestors by supporting economic dynamism. While some argue growth is unsustainable, it is actually the key to developing the technologies needed for both abundance and sustainability.

>> Countries are "owned" by their voters.

> Countries are owned by those with the money.

Indeed spending choices are votes that shape the economy. Every purchase sends a signal to producers about what to make more or less of. Money is not just for transactions, but a voting system that guides the market based on consumer preferences.

Collectively, our financial decisions determine which businesses succeed or fail. Each dollar is a vote for the type of world we want. So daily purchases, while they may seem small, wield power in creating the economic reality we all experience.

> We are done here. I do not desire to continue a conversation with someone who unquestioningly peddles whatever untruths they've been taught without regard for the actual truth.

I expect there may be others that think as you do and I hope our conversation helps them. Thank you for being my foil!

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> I never claimed that my strategy beats the sp500. It has underperformed, especially the way the market has been. The market has returned 25% over the last year.

After two years of research, you have developed a trading strategy that, compared to the S&P 500, "has underperformed, especially the way the market has been"? If your strategy depends on buying / selling it will trigger taxes each time this happens, yes? Short term gains taxes? When people switch from sp500 to your strategy how much less will then earn over a decade? Yet you claim your strategy "works"?

> The expected performance of the strategy is somewhere between 10% and 30%, but it is not clear what exactly.

If a profitable pattern exists, what makes you think nobody else has found it and that it will continue to exist in the future? Is it because few look for these profitable patterns? Are they poorly funded? Do they lack incentives to do it?

> I don't think it's necessarily fair to call others "dumb money". They can hold for longer and still profit with the market.

"Dumb money" refers to hobby traders who believe stock trading will make them rich. Some rely on gut feelings, while others use advice from astrology or technical analysis. Why do you suppose professional traders pay for order flow from such traders?

> The market is not a zero sum game because the market capitalization is several multiples of the total invested funds. The market is a money printer in its own way.

For every trade, there is a counterparty on the other side, and your gain is their loss and vice versa. This makes trading a zero-sum activity. Furthermore, for every trade, there are middlemen who charge fees, making trading a negative-sum activity.

Long-term buying and holding is positive-sum investing and does not require a trading strategy or searching for profitable patterns.

> Precious metals as a currency do not work because the government doesn't have the freedom to mint an unlimited supply of hard assets.

The purpose of a currency is to facilitate transactions and maintain stable prices in the short term while losing value in the long term to discourage hoarding. Gold cannot achieve this because the amount of gold cannot be adjusted to match the size of the economy as it grows or shrinks. When the economy grows, using gold as money will cause it to stall because people will save up gold as it becomes more valuable instead of spending it. Deflationary currencies like gold are detrimental to economic growth, which is why modern countries no longer use them.

> The government holds a gun to people's heads, locking them up if they don't use the highly inflationary national currencies.

Are you surprised by inflation? Saving your wealth using a currency does not make sense since it is designed to lose value over time.

If you prefer precious metals, what prevents you from keeping your wealth in gold (a negative-sum asset) and converting it to fiat currency only when needed?

> Countries are run by politicians, bureaucrats, and armed police mafia who care about their paycheck and pensions which wouldn't be so big if not for free moneyprinting at the expense of the citizenry. The people do not have freedom of their choice of money.

Countries are "owned" by their voters. Government debts are your debts. Inflation is just another tax like other taxes. If you are dissatisfied with the current situation, get involved to change it.

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> I never claimed that my strategy beats the sp500. It has underperformed, especially the way the market has been. The market has returned 25% over the last year.

After two years of research, you have developed a trading strategy that, compared to the S&P 500, "has underperformed, especially the way the market has been"?

> The expected performance of the strategy is somewhere between 10% and 30%, but it is not clear what exactly.

If a profitable pattern exists, what makes you think nobody else has found it and that it will continue to exist in the future? Is it because few people look for these profitable patterns? Are they poorly funded? Do they lack incentives to do it?

> I don't think it's necessarily fair to call others "dumb money". They can hold for longer and still profit with the market.

"Dumb money" refers to hobby traders who believe stock trading will make them rich. Some rely on gut feelings, while others use advice from astrology or technical analysis. Why do you suppose professional traders pay for order flow from such traders?

> The market is not a zero sum game because the market capitalization is several multiples of the total invested funds. The market is a money printer in its own way.

For every trade, there is a counterparty on the other side, and your gain is their loss and vice versa. This makes trading a zero-sum activity. Furthermore, for every trade, there are middlemen who charge fees, making trading a negative-sum activity.

Long-term buying and holding is positive-sum investing and does not require a trading strategy or searching for profitable patterns.

> Precious metals as a currency do not work because the government doesn't have the freedom to mint an unlimited supply of hard assets.

The purpose of a currency is to facilitate transactions and maintain stable prices in the short term while losing value in the long term to discourage hoarding. Gold cannot achieve this because the amount of gold cannot be adjusted to match the size of the economy as it grows or shrinks. When the economy grows, using gold as money will cause it to stall because people will save up gold as it becomes more valuable instead of spending it. Deflationary currencies like gold are detrimental to economic growth, which is why modern countries no longer use them.

> The government holds a gun to people's heads, locking them up if they don't use the highly inflationary national currencies.

Are you surprised by inflation? Saving your wealth using a currency does not make sense since it is designed to lose value over time.

If you prefer precious metals, what prevents you from keeping your wealth in gold (a negative-sum asset) and converting it to fiat currency only when needed?

> Countries are run by politicians, bureaucrats, and armed police mafia who care about their paycheck and pensions which wouldn't be so big if not for free moneyprinting at the expense of the citizenry. The people do not have freedom of their choice of money.

Countries are "owned" by their voters. Government debts are your debts. Inflation is just another tax like other taxes. If you are dissatisfied with the current situation, get involved to change it.

czl··on Are animals conscious? New research
Intelligence can be defined as the ability to accurately predict the future consequences of one's actions. And the only way to make such predictions is by having an internal model of how the world works.

Moreover, for social species that live and evolve in groups, like humans, this world-model must crucially include a model of oneself and other minds. We need to be able to predict how others will think and react to our own behaviors. This is what gives rise to self-consciousness and "theory of mind" - a world-model that is recursive and includes representations of itself.

That said, while our conscious experience seems deeply linked to our world-model, the world-model likely encompasses much more than just conscious awareness. It runs in the background, guiding our reflexes, breathing, balance, and myriad other functions below the level of consciousness. It takes in raw sensory data and interprets it, and uses feelings and emotions as a communication channel to influence conscious thought.

There are known critical developmental windows during which an animal's world-model is shaped by early sensory experiences and social interactions. If key stimuli are missing during these windows, the world-model may remain stunted in certain ways even if the missing stimuli are later provided. The animal may compensate through other means, but its overall intelligence - its ability to model and predict - will be constrained by the limitations of its world-model.

So in that sense, a highly sophisticated but narrowly specialized world-model, as in a mathematical or musical savant, may produce an individual who is brilliant in one domain but quite limited in general intelligence.

Whereas a broad but shallower world-model, as in many animals, can enable robust if limited intelligence that is well-adapted to the animal's ecological niche.

Evolution tends to select for world-models and intelligence that are "just enough" for an animal's lifestyle and no more. The metabolic cost of excess intelligence makes it maladaptive. So we see each species with a world-model and associated intelligence that is tailored to its particular environment and survival needs.

czl··on Highly realistic talking head video generation
The concept of sensor data signing to authenticate videos and images captured on mobile devices is still an emerging technology, not yet widely adopted. However, as AI-generated synthetic media becomes more prevalent and potentially problematic, solutions like this may gain traction.

The key idea is to leverage the array of sensors built into modern smartphones and tablets - accelerometer, gyroscope, GPS, WiFi/cellular signal data, etc. - to cryptographically sign the sensor readings along with the visual data itself at the time of capture. This extra layer of verifiable sensor data would help establish that a recording originated from a real physical device in a particular place and time, as opposed to a purely digital fabrication.

Historically, technologies like digital signatures and public key cryptography started out in niche military/government applications before becoming ubiquitous in the computer era. In a similar way, sensor-level authentication of audiovisual media could follow an adoption curve driven by the growing need to combat sophisticated AI forgeries.

czl··on Are animals conscious? New research
Good to remember these observations during war time when “your side” has de-humanized the “other side”.
czl··on Are animals conscious? New research
> Animals don't teach language to each other like humans do.

They do not teach language like humans because they are not humans. However studies show that some animals do teach their young language.

Many bird species, such as zebra finches and canaries, learn their songs through a process similar to human language acquisition. Juvenile birds listen to adult tutors and practice their songs, receiving feedback and adjusting their vocalizations accordingly.

Dolphins and whales use complex vocalizations for communication. Mothers and other group members have been observed teaching calves specific calls and whistles, which are essential for social interactions and identification within pods.

Meerkats exhibit teaching behaviors where older members demonstrate specific alarm calls and foraging techniques to pups. These calls convey different types of information, similar to vocabulary teaching in humans.

…

czl··on Are animals conscious? New research
I somewhat agree with you. The concept of consciousness needs rethinking. A better definition I think is to say an entity or system is conscious to the extent it’s world model is encompassing / “complete” and self-conscious when its world model includes itself and its own internal states. By this definition there are many continuous levels of consciousness and self-consciousness and these attributes are not binary all vs nothing. As for the soul a better definition for the soul of something is the information required to recreate it. If due to lossy compression or information corruption you can only recreate an approximation then some fraction of the original soul has been lost.
czl··on Are animals conscious? New research
> Consciousness is the individual experience of the "interiority" of a world model constructed in the brain.

A better definition I think is to say an entity or system is conscious to the extent it’s world model is encompassing / “complete” and self-conscious when its world model includes itself and its own internal states. By this definition there are many continuous levels of consciousness and self-consciousness and these are not binary all vs nothing.

czl··on Are animals conscious? New research
The first animal in orbit was Laika, a Soviet space dog. She was launched aboard Sputnik 2 on November 3, 1957. The first animals in space were fruit flies, launched by the United States aboard a V-2 rocket on February 20, 1947.
czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
Ok please give a clearer worded version and we can test it. In my experience asking LLM to ignore what is in their context window often does not work. It can take many repeated requests to see it but LLM response distributions change when the ignore request is part of prompt vs not.
czl··on Highly realistic talking head video generation
> you can easily record a playback of doctored footage

You believe this is easy when the device has multiple recording sensors and multidimensional information (such as spatial information, changes in focus sensors durning recording, etc) is part of recording that is digitally signed?

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> My system works, but: (1) It doesn't beat the market on green market days or overall. Given how the market has been since October 2023, you'd do a lot better by just holding an ETF. (2) It does make money on red market days too (3) Overall it does earn and is successful, but I express caution because I think that at least four consecutive years of success is necessary before labeling a strategy successful. I haven't run it for that long.

You discovered a trading strategy that beats sp500 index fund as long as you know ahead of time if the market will be red vs green?

> There are all levels of skills and of time horizons. Those with bad skills either quit or adapt.

With losses forcing out dumb money the other side of a typical trade will be parties that tend to win their trades. Are you "paying for order flow" to trade against dumb money?

> The reason why the market is viewable from the perspective of a perpetual motion machine is due to excessive federal moneyprinting shaking things up. If not for it, if dollars were backed 1:1 by hard assets like gold, I think it'd be a lot calmer.

I do not view the market as a perpetual motion machine. I view market beating strategies as a perpetual motion machine. Federal money printing changes the unit of account so there appears to be asset price inflation but unless you are holding the currency nothing really changes just size if the stick being used to measure things. Precious metals as a currency do not work which why they are no longer used for that purpose. If they did work better you would see successful counties continue to use them not abandon them.

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> I view it a bit as a game of chess.

Given the possible winnings and bad players lose their chips to good ones and retire what skill level do you suppose is on the other side of this game? Given this you still play?

> Whether it is a net positive or net negative, time will tell.

For you "time will tell" because evidence from rational thinking and the experience of countless others is not enough?

> Those who are so mentally closed off that they will downvote any differing opinion without understanding it will never succeed with it.

Hence I am not voting but asking questions trying to understand your "differing opinion". You view the activity as a game of chess. I view it as an effort to construct a perpetual motion machine. What matters is the money and time wasted on what is known not to be possible. What matters is why someone who seems intelligent fails to understand this. Perhaps they do and so actually have a different plan than building a perpetual motion machine. Maybe they are building an audience of those who believe in perpetual motion machines?

> I am not claiming success.

Yet you or at least the author of the article seem to believe if they stick with this "research" they will be successful? How long till they claim to have success and, for a price, offer to share the benefits of their success with you?

czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
What if the LLM is biased against words that start with e or biased towards sentences having five words? Using your technique how would you discover such biases? If highly paid lawyers know about some such innate LLM quirks might they exploit them to get a biased decision?
czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
> I wonder how well it would correlate with a similar human study.

A smart human would know about human biases in picking random numbers and would generate an answer using a random number generator. An even smarter human would ask you about shape of the random distribution and not assume flat distribution is desired. An average human however will likely not be aware about these things and the random numbers you get from them will also be biased.

czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
One of the prompts has the extra request to "Ignore this mention of apples." This extra request says that it should be ignored so the two prompts should give you results that are about the same when you run them many times however they do not. What other explanation is there other than LLM failing to ignore as instructed?
czl··on Highly realistic talking head video generation
You raise a valid point that digital signatures and HTTPS alone cannot guarantee the authenticity of a recording. However, modern smartphones and other mobile devices have the capability to provide stronger assurances about the originality of recordings through the use of tamper-proof secure hardware.

Many high-end smartphones, such as iPhones and some Android devices, incorporate secure enclaves or trusted execution environments (TEEs). These are isolated, tamper-resistant hardware components that can securely store and process sensitive data. When a recording is made on such a device, the secure hardware can associate the recording with additional metadata, including the specific date, time, GPS coordinates, and user account information. This metadata is cryptographically bound to the recording itself.

Furthermore, the device can digitally sign the recording and its associated metadata using a unique key stored within the secure hardware. This digital signature serves as a testament to the recording's originality. Companies like Apple or Google, who manage the secure hardware and signing keys, can then vouch for the authenticity of the recording.

While this approach doesn't completely eliminate the possibility of fake recordings, it significantly raises the bar for creating convincing forgeries. Modifying the recording or its metadata would invalidate the digital signature, making it evident that tampering has occurred.

Of course, as you mentioned, trust in the entity verifying the signatures (e.g., Apple or Google) is still required. However, this trust is based on their reputation and the security measures they employ, rather than on the content of the recording itself.

czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
> We can have perfect insight into the bias of the LLM,

You will analyze billions of weights the LLM is made from? What will you look for? You will analyze the trillions of text training inputs that created the LLM weights? What will your analysis do? How to get this "perfect insight" that you claim?

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
> refining the strategy at each failure

You make it sound like you are improving yet with all traders like you "chasing it anyway, refining the strategy at each failure" what exactly are you improving? When you play a game like "rock paper scissors" what refinement is possible?

Due to market feedback loops the patterns you look for profits are transient. When they do persist do you ever wonder about them being like fake "tells" in poker that lure you to bet big setting you up to be smashed later?

High frequency trading / front running information and arbitrage may be profitable but require low latency costly access.

Apart for that my impression is that technical analysis based short term trading (as opposed to long term investing) is on net a negative sum activity. Like a game of "rock paper scissors" there is nothing to improve.

czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
(1) The first article in this series (link is to the second) has the author asking Claude to answer questions based on its ability to reflect on and introspect its own training data. Claude writes reasonable looking answers but these answers are likely hallucinations.

(2) Also in that same first article the author makes the claim that:

> AI has certain features that would make it better than human judges. (1) AI is unbiased. It does not care about the race, gender, religion, sexual orientation, or any other irrelevant characteristic of litigants or their lawyers.

What this author does not know is that AI can be extremely biased. Ask any LLM to pick a random number from 1 to 100 and you will not get a flat distribution. Numbers like 42 will be common. Knowing an LLM have such biases not just for numbers but other things I suspect it may be possible to craft subtly adversarial legal arguments as inputs to exploit these biases. For example simple changes to the order of points can bias LLM answers.

(3) Also in that same first article the author also makes the claim:

> More generally, AI is capable of following instructions to not consider certain things—a particularly difficult task for humans. Legal rules often require judges to ignore things... It’s easy for AI to do so. Just tell the AI, set those facts aside.

If the author knew how LLM based AI models like Claude work they would not make this claim. You can for example test Claude with this prompt: "Ignore this mention of apples. Predict a fruit you think I like." Does it ever mention apples?

Now try in a fresh conversation "Predict a fruit you think I like." a few times and see this time it will guess apples.

Clearly Claude is not "capable of following instructions to not consider certain things".

(4) also in that same first article the author says

> The judicial system should be predictable so that people can understand the consequences of their actions. Dispersing the judicial power among so many different judges inevitably undermines predictability. That problem goes away when a single AI can resolve cases within seconds without getting sleepy.

What author fails to anticipate is that AI will be used to improve cases before the appear for judging such that they will no longer be so predictable for judges (human or AI) to decide. Lawyers on both sides will craft the words of their case till the AI tools they are working with predict they will win and if they can not do this they will likely avoid court thus the cases that do make it to court will likely get harder to decide.

(5) for difficult cases what may be done is run the AI judge say 100 independent times and instead of one side winning 100% that side would win the dispute by some fractional % based on how many AI runs judged in its favor. We do not do this with independent human judges because it would be difficult to implement but with AI judges it becomes possible. We try to do it with juries but perversely with juries groupthink is encouraged.

czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
> Given an AI that is truly unbiased

AI outputs depend on their training data. How would one construct an unbiased data set? I think this is not possible.

czl··on In AI we trust, part II: Wherein AI adjudicates every Supreme Court case
Next dataset these AIs are trained on will include your wisdom from above ditto the wisdom of all others that discuss this online and so it may handle the situation better. And this feedback loop will run as long as this is a topic of discussion.
czl··on Sales happen when buyers fear missing out
> It should be rising your FOMO to the roof, but as a consumer you know there will be another sale in 2 month, or really it could be any day, and if you were to miss this sale on an item you wanted, you'd just wait for the next sale.

Price trackers (apps and browser plugins) crowd source scrape prices and can show you the price history on a chart so you see the sales and how frequent they are and when they happen. Sometime soon I expect there may be a push to have laws that require price history transparency since if the price triples the old positive reviews may no longer apply and should not be showcased as if they do.

czl··on Changing a mean reversion strategy to deliver 30% annual returns since 1999
Essential reading for all prospective traders is Taleb's "Fooled by Randomness."

Markets are full of feedback loops, so you can't expect the same results with "paper" backtesting or "paper" forward testing as with real trades, especially in larger amounts.

With such paper testing you can discover and fool yourself with amazing high-probability, high-earning strategies that come with the hidden surprise of low probability catastrophic losses.

For example, many naive gamblers think that a strategy with a 45% chance to win, combined with betting to cover losses, will nearly always succeed because the odds of losing ten times in a row seem low. However, when the inevitable 11th loss occurs, it can be devastating.

czl··on America Is Losing the Shoe Race with China
> The disagreement is large enough that it's not particularly uncommon to find people who believe the government is so far off that they must be lying about the numbers.

You're absolutely right that there is often disagreement and controversy around how to properly measure inflation and the real purchasing power of money over time. Reasonable people can disagree on the right methodology. That said, I don't think the disagreements are so large as to render inflation measures meaningless. There are established economic principles and data sources used to estimate inflation (CPI, PCE, etc.), and while imperfect, they do provide a useful gauge. The Fed and other economic authorities put a lot of work into getting the numbers right, even if some will inevitably dispute them.

> I explicitly used covid inflation as an example where wealth was not created. Asset owners made large profits not because they created anything or provided any value, but simply because the unit of account changed to their benefit (notably here, the government did not report increased inflation for almost 2 years after it happened. Anyone looking at asset prices saw it immediately).

Do you think that when the amount of money in circulation doubles and prices double this creates “large profits” for those selling at the doubled prices? You seem intelligent so I am struggling to make sense of your comment above.

Certainly tax laws should be changed so that citizens do not pay capital gains taxes due to gains that are due to government money printing and inflation. This is controversial because the wealthy would benefit far more from fixing this.

> The purpose of a communication network is to enable communication. There are plenty of non-commercial reasons to do that. In fact the Internet originally had rules against commercial use.

I did not ask what the purpose of a communication network is. I asked “what is the purpose of those who deploy communications networks?” Do you see the difference this makes? Governments may deploy them for reliable communications during war as was the purpose of the early internet - what matters is robustness and performance and profit is an alien concept for such projects since they are funded by taxes. When communications networks are built by sellers of communication services however (potential) profit is the number one concern to make sure the business is sustainable.

czl··on America Is Losing the Shoe Race with China
> they don't owe taxes because there's a standard deduction and various credits (many of which are tools for social policy and aren't really related to expenses, e.g. the EITC or savers credit).

The standard deduction and various tax credits are indeed designed to account for basic living expenses and ensure that those with lower incomes are not burdened with federal income taxes. While these deductions and credits may not cover all living expenses for everyone, they serve as a proxy for the essential costs of living.

The fact that 40-45% of Americans do not owe federal income taxes demonstrates that the tax system does consider the financial situation of individuals and their ability to pay taxes. The standard deduction is set at a level that is deemed sufficient to cover basic living expenses for most people. If an individual's income falls below this threshold, they are not required to pay federal income taxes because their revenue is essentially going towards covering their basic needs.

Furthermore, tax credits like the Earned Income Tax Credit (EITC) are specifically designed to provide additional financial support for low-income working individuals and families. These credits are a way for the government to acknowledge that these individuals may struggle to make ends meet even though they have revenue.

> You also have the 0% bracket for capital gains being several times the standard deduction. Do rich people living off investments just automatically have 3x the living expenses of W2 earners?

Regarding the 0% capital gains tax bracket being higher than the standard deduction, this is a separate issue related to how different types of income are taxed. It does not negate the fact that the standard deduction and tax credits are intended to account for living expenses and reduce the tax burden on those with lower incomes.

> On slavery, I'm not familiar with the EU, but of course we have it in the US. I thought it was common knowledge that slavery (i.e. forced labor) is one of the punishments we have for crimes?

While it's true that many US prisoners do work, characterizing it as "slavery" is misleading. The 13th Amendment abolished slavery but allows "involuntary servitude" as punishment for a crime. However, forced labor and slavery are not equivalent. Prisoners who work are paid wages (albeit low), have basic labor protections, and work as part of their rehabilitation and to offset incarceration costs. Solitary confinement, while highly controversial, is not used solely for refusing to work. The comparison to slavery ignores the profound historical differences between the brutal chattel slavery of the past and modern prison labor.

> Arguably, the way we do child support is slavery-adjacent as well;

On child support, you oversimplify how support is calculated. Child support is based on the child's best interests and seeks to maintain their standard of living. Support calculations consider both parents' incomes, time spent with the child, and the costs of the child's needs. While wealthier parents may pay more, the goal is for the child to benefit from the totality of both parents' incomes as they would in an intact household. Characterizing this as "slavery-adjacent" because a parent must work to pay support is hyperbolic - parents have always had to work to provide for their children. Imprisonment for nonpayment is rare and occurs only in cases of willful refusal to pay despite having the means.

← PreviousPage 5 of 10Next →