Hacker News Reader

iciacon The efficient frontier of LLM inference

The authors are referring to a Pareto Frontier https://en.wikipedia.org/wiki/Pareto_front

iciacon Show HN: Crossing The Lexicon, a fun Wordle-inspired daily word game

Great game! Curious what design decisions you would change if you were starting from scratch.

iciacon The 'attention economy' corrupts science

“What information consumes is rather obvious: it consumes the attention of its recipients. Hence a wealth of information creates a poverty of attention, and a need to allocate that attention efficiently among the overabundance of information sources that might consume it.” - Herbert Simon

iciacon UST Stablecoin Loses Dollar Peg

111TWh is well beyond a typical hydroelectric dam output (most are an order of magnitude lower storage and generation). As an example, the very sizable Australian Snowy Hydro 2.0 upon completion is estimated to have storage for 350GWh (the Australian national energy market is ~190TWh). The original Snowy Hydro (9 stations) has annual energy production of ~5GWh.

iciacon Ask HN: Which books in your field do you think are perfect for self study?

There are two out-of-field books that I always recommend to policy analysts, economists, and regulatory drafters: The Design of Everyday Things by Don Norman, and Algorithms To Live By by Brian Christian and Thomas Griffiths. Both are high signal-to-noise primers on topics that are relevant in decision making and policy, but are rarely covered in an economics or public policy curriculum.

iciacon Ask HN: Which books in your field do you think are perfect for self study?

Equally "The Art of Doing Science and Engineering" by Hamming is one of the best books around on the philosophy of problem solving, and an excellent primer on core concepts in signals processing, information theory, and computing.

iciacon Elegant six-page proof reveals the emergence of random structure

You might be interested in this 1 page paper by John Nash, which proves the existence of equilibria for finite N-player games (an extremely powerful result). In essence it uses a set theory result (Kakutani's fixed point theorem), and simply notes that his description of a N-player game meets the required conditions for that result to hold. http://www.sscnet.ucla.edu/polisci/faculty/chwe/austen/nash1...

iciacon Braess Paradox

Braess Paradox is one of the nicest results in network science and traffic engineering. It's possible to find examples of its application in any situation that can be modelled as a network, from graph neural network architecture (removing connections and inducing sparsity can lead to better generalisation and efficiency under conditions); economics (introducing new trade connections can reduce wellbeing and efficiency under conditions); organisational theory (introducing firewalls between teams can reduce the prevalence of groupthink); and many more.

iciacon Braess Paradox

There's an analogous concept to Nash equilibria in transport engineering known as Wardrop's First Principle. In essence, at equilibrium no user has an incentive to change their behaviour by choosing an alternative route. A 'central routing algorithm' that optimises over the system is in essence Wardrop's Second Principle. https://en.wikipedia.org/wiki/John_Glen_Wardrop

iciacon Backyard approaching lighting at Adelaide Airport

As a related Adelaide fact - the center of town is ringed by a "moat" of parklands, each ostensibly the width of a cannonball and designed as a defensive structure (an invading force would need to run through a cannon's worth of artillary). On a map the green square is extremely distinctive https://en.wikipedia.org/wiki/Adelaide_Park_Lands#/media/Fil...

iciacon Researchers can steal data during homomorphic encryption

As an aside it's worth noting that RSA itself is partially-homomorphic (ciphertext multiplications are preserved in the decrypted plaintext).

The idea of 'homomorphic encryption' was even introduced by another Rivest and Adleman paper, almost immediately after the famous 1977 RSA algorithm ("On Data Banks and Privacy Homomorphisms" by Rivest, Adleman, and Dertouzos 1978).

iciacon Star Tek–Exploiting the Final Frontier: Counterspace Operations in 2025 (1996) [pdf]

This is a genuinely fascinating read. Thank you for sharing.

iciacon Drug cartel assassinates its enemies with bomb-toting drones

https://www.cato.org/publications/policy-analysis/how-legali...

Cato report (2018) on the effects of marijuana legalization on border smuggling. It's had a large and direct impact on the market.

iciacon Universal Basic Income is Capitalism 2.0

To correct this slightly, a carbon tax changes the relative price of goods (rather than a consumption tax, which is typically flat). The primary industries affected tend to be industrial, rather than than consumption based (e.g. energy, freight, mining, and smelting).

iciacon A deep learning framework for neuroscience

Learning's likely to be bidirectional. ANN (as a mathematical analogue) is independent to the biological function (the original and key inspiration). Advances in network architecture (e.g. the recent trend towards skip connections and parallel processes) is likely to give insight to how an underlying, more complex system is likely to operate. In particular, systematic errors made by ANNs under given frameworks have a tendency for existing in some form in psychology and biology. Since conceptual thinking from both domains can directly feed towards each other, it's a rare bootstrap moment with the potential for rapid advances in both directions.

iciacon Australia Will Join U.S. for 2024 Moon Mission, Mars Exploration

Currently, the focus of the organisation has been on developing Governance, MOU's etc. $10M is basically seed to establish administration.

iciacon The Good Employee, a story about how you can explain companies with graph theory

Best to read this blog as an example of economic behaviour explained through a graph - valuable in itself, as even basic graph theory concepts rarely make it onto an economics curriculum. Which is a shame, given how neatly algorithmic thinking and complexity costs improve a standard rational agent model (e.g. a few behavioural economic concepts, such as myopic discounting, pop out naturally if you assume mental costs to imagining future states).

Conceptually, it's interesting and potentially foreign to the target reader - more formal definitions can wait for an academic paper rather than a casual blog.

iciacon What Minimum-Wage Foes Got Wrong About Seattle

Wage distribution certainly does not resemble a bell-curve. It's a long-tail distribution. This is observed consistently across different economies. This US figure is from 2015, however it should give the right idea:

https://www.census.gov/library/visualizations/2015/demo/dist...

iciacon Tech companies pay poor Kenyans to produce training data for AI

It's not only developing countries that benefit from increased female participation in the labour force. For Western nations, the increased proportion of women in the labour force was a core component of the rapid increased economic growth between the 1940's and today.

Low fertility is an interesting thing where it comes to development; for most countries it has quite dramatically declined as incomes have increased. The effects are likely bidirectional (increased economic growth -> lower fertility rate; lower fertility rate -> increased per capita income).

The drivers are complex and interrelated: high child mortality rates, low life expectancy, and high poverty are associated with high fertility. It's an unjust comparison to make, however leading theories share common features with K-r reproduction strategies in ecology (with increased survival / longevity, investing in a few highly educated/skilled children becomes feasible). Similarly, increased education / maternal health is highly correlated with economic development and child wellbeing. These are often highly compounding effects over time - a virtuous cycle in which increased economic freedoms, health, education, and opportunities (male and female - often females are relatively impoverished, so there's the potential for bigger immediate benefits) can lead to rapid development with the right institutional conditions.

iciacon What Minimum-Wage Foes Got Wrong About Seattle

Automation usually requires a high fixed cost to enable long-term low-operating costs. If you suppose a decreasing unit installation cost or scale construction over time you'll eventually hit a point at which companies will automate, especially if there's added efficiency (two examples: McDonald's certainly didn't shift to kiosk solely due to minimum wage, nor could we set an hourly wage low-enough that a typist pool could compete with a common word processor).

Ditto for location shifting: it's dependent on a) the transferability of the activity, b) the cost of shifting, and c) the price differential between the two locations. Minimum wage may be a component in this decision - but it will not be the only one.

For context: I work as a policy economist in Australia, principally in regional employment. The minimum wage rate here is (AU) $18.93/h, unemployment sits at trend around 5.5%. In my experience the primary driver for regional employment (often less complete markets) is labour demand rather than the cost of supply.

iciacon What Minimum-Wage Foes Got Wrong About Seattle

The theory of the minimum wage is just the theory of a price floor. The effect on employment is dependent on the market composition and the cross-price elasticity of labour. Comparing a price floor of $15/h (highly substitutable unskilled labour) with $1000/h (unsubstitutable specialist labour) isn't really tenable.

For most minimum wage positions you'll find the bulk is employment by large firms (potentially under a franchise arrangement), labour is substitutable, and wage-bargaining power is weak. Total production is more likely to be driven by demand, and a (relatively) small unit cost increase is likely to be absorbed (if anything, we'd expect it to affect firm profit more than purchased labour).

You're correct that less competitive SMEs may be affected more here, operating as a price taker for both their product and labour. These firms however will be disproportionately affected by any external market movement - there's nothing particularly magical about a minimum wage increase.

iciacon What Minimum-Wage Foes Got Wrong About Seattle

Wages can certainly track lower than marginal productivity. That's the expectation under monopsonistic labour markets (few or one buyers of labour, many substitutable potential employees). The observed employment effects in response to the minimum wage increase here is pretty well standard theory. https://en.wikipedia.org/wiki/Monopsony#Static_monopsony_in_...