HNHacker News
TopNewBestAskShowJobs

loveparade

2,076 karma · joined December 30, 2022

Meet Me At The Love Parade
submissionscomments
loveparade··on John Carmack’s ‘Different Path’ to Artificial General Intelligence
I'm always taken aback by this fascination and heroification with John Carmack (or similar figures) on HN. I love Doom and Quake too, but that doesn't qualify him any more than any other random senior engineer to work on something totally different.

From his interviews it looks like he understands little about the technical details of ML, or about as much as anyone can learn in a few months, and is just banking on PR based on his games and name.

I put him into the same category as Elon Musk, who also understand nothing about technical details of AI, but was still able to hire a world class team at OpenAI. His name and fame counts for something in terms of recruiting and joining his venture may be a good bet because of that, but he's not a person whose opinion on the subject matter I would take seriously in the same way I'd take a researcher seriously.

loveparade··on Alphabet’s Profit Falls 34% (fourth consecutive drop) Amid Ads Slowdown
I don't think it's that simple of a hypothesis to verify, even inside Google. There are lots of reasons why CTR may be constant or go up despite fewer people converting through ads. Examples are bots, crawlers, more kids playing with iPads, or malicious website design, e.g, I have clicked on ads because I was tricked on it by the HTML/JS, but that doesn't help the advertiser because I'm not going to buy anything.

CTR only matters in the short-term. In the long-term, what matters is if the market of advertisers believes ads are profitable for them. If their CTR goes up 10x but they cannot verify conversions, they may eventually stop buying ads. That won't happen quickly, because attribution is so hard and you can never be sure where your conversions really come from, but the decline of ads may still happen over a longer time frame.

loveparade··on Alphabet’s Profit Falls 34% (fourth consecutive drop) Amid Ads Slowdown
Yes, but CTR and conversion metrics still matter to ad buyers. It's much more difficult to measure the influence on my subconscious and attribute that.
loveparade··on Alphabet’s Profit Falls 34% (fourth consecutive drop) Amid Ads Slowdown
The fact that the ad model has worked so well is always a nice reminder of how much of a bubble I live in. Personally I have never intentionally clicked on a Google/YouTube/etc ad. Never. I wonder if the general population has been getting more tech-savvy and skeptical of ads over time, but this change takes a very long time to play out. I sometimes see my dad clicking on ads because he doesn't understand the mechanics behind them, but he'll die pretty soon, and a younger more tech-savvy generation will replace him.
loveparade··on Tether ownership and company weaknesses revealed in documents
Many retail investors probably have. But the big Tether holders can't liquidate their large positions because it would likely cause a market collapse. Many holdings are on public addresses, so if for example a big exchange liquidated their Tether, everyone would know and start panic selling, leading more big holders to sell, and so on. And even if they liquidated through OTC or centralized parties that are not publicly known, the volume would still obvious.

Hence it's in the best interest of many parties to keep Tether up to avoid a market collapse.

loveparade··on Release engineering is exhausting so here's cargo-dist
It's interesting how many people comment on nix without seeming to know anything about it. Why?

Nix doesn't replace cargo. It would use cargo under the hood. If Cargo supports Windows builds, those builds will work with nix. However, nix (probably) doesn't give you anything additional in this case.

loveparade··on Release engineering is exhausting so here's cargo-dist
Nix builds binaries. You don't need nix to run binaries built using nix. Just like you don't need Cargo to run binaries built using cargo. This is about developer tooling.
loveparade··on Release engineering is exhausting so here's cargo-dist
Nearly all of these problems (except for maybe the Windows one) are solved by publishing a nix flake with your application and telling people to install that instead.

Not everyone uses nix, but I'd rather push its adoption than trying to build these kind of solutions for all possible language ecosystems when a general solution already exists and works great.

loveparade··on Ask HN: How do you exercise?
I do a combination of Crossfit and olympic weightlifting 4-5x per week. I usually do my olympic lifts before Crossfit so that I'm not too tired for them. That means I don't have full energy for the WODs, but that's totally fine. I also do some mobility work for 15-30min each morning.

So many people seem to hate Crossfit, and while some of the criticism is valid, it has been the only thing that was easy to stick with for me, mostly due to the community. Many previous attempts of going to the gym alone regularly failed after the motivation faded a few weeks into it. For all it's flaws, I still recommend it to newcomers. The best program is the one you stick with.

I started adding more olympic weighlifting practice 1-2 years after starting Crossfit because I felt like I didn't get enough practice for the more technically difficult lifts like snatches from Crossfit alone.

loveparade··on There’s no such thing as affordable housing
> High quality goods? Too expensive to produce and lasts too long to be economically viable.

I think that's a false dichotomy. Free markets typically produce higher-quality goods and services than non-free markets because quality is part of the dimensions among which firms compete. Throughout history, quality of goods in all kinds of areas have gone up when government restrictions were loosened and competition was allowed.

Preventing a free market in housing benefits exactly one group: Those who are lucky/rich to already have such a nice big house.

loveparade··on There’s no such thing as affordable housing
That's the ever-repeating story of the disconnect between economics and politics. From an economic optimization standpoint it's pretty obvious what needs to happen, but those things will be painful for a subset of citizens, which means less votes for the next election, which means no politicians want to do it.

That's not just the case for housing, but a whole range of regulations, many of which only still exist because removing them will lead to voters becoming upset about short-term changes, even though it may benefit the economy in the long run.

loveparade··on Ask HN: ML Papers to Implement
I believe which paper you implement matters less than you think. Most papers take some kind of well-known base model, make a few tweaks, and then run experiments. To implement it you need to follow the references and start with the base model, and 99% of all papers within some field will eventually lead you back to implementing the same base model, so it really doesn't matter where you start. The overlap is huge.
loveparade··on Ask HN: ML Papers to Implement
> ideally, a list of papers that could take 2-5 hours each and a few hundred lines of code?

I think you are severely underestimating the time required, unless you are quite experienced, know exactly what to look for, or the paper is just a slight variation on previous work that you are already familiar with.

Even seasoned researchers can easily spend 30+ hours on trying to reproduce a paper, because papers almost never contain all the details that went into the experiments. You are left with a lot of fiddling and iteration. Of course, if you only care about roughly reproducing what the authors did, and don't care about getting the same results, the time can be much shorter. If the code is available that's even better, but looking at it is cheating since wrestling with issues yourself is a big part of the learning process.

A few people here mentioned Andrej's lectures, and I also think they are amazing, but they are not a replacement for getting stuck and solving problems yourself. You can easily watch these lectures and think "I get it!" because everything is so well explained, but you'll probably still be stuck when you run into your own problems trying to reproduce papers from scratch. There's no replacement for the experience you gain by struggling :)

It's like watching a math lecture and thinking you get it, but then getting stuck at the exercise problems. The real learning happens when you force yourself to struggle through the exercises.

loveparade··on Is Crypto Dead?
There was little crypto fraud this year, other than the usual hacks. All the big stories were good old regular fraud by centralized companies that you also see in the fintech space. None of this was enabled by crypto. The problem is that most people don't understand the difference. It was just fraud, not crypto fraud. It's like saying that Theranos is "blood test fraud" and thus we should ban all blood testing. The fact that some founder scammed customers and investors has nothing to do with the market.
loveparade··on Ask HN: Why do games (as media) make so much money?
> Most big spenders are incredibly wealthy, like buy $30,000 wristwatches for fun wealthy. Games have figured out how to 'go infinite' on their spend possibilities and capture wristwatch money from these folks.

Do you have any data to back up this claim? I don't have data to prove the opposite either, but I don't believe this is true. I know multiple big spenders in real life. They have a very average income, but instead of putting the ~$500-1000 that's left each month into a saving accounts, they spend it on mobile games, i.e. make bad financial decisions. FOMO is probably the main driver.

I'm sure the people you refer to exist. There are whales that are so rich that they don't care about wasting $10-100k on a game, but I don't believe these are the majority of players that spend large amounts. I think the big spenders on average have below average income, they are just bad at self-restraint, and/or relatively young and inexperienced with money.

loveparade··on Ask HN: Why do games (as media) make so much money?
That's true for Southeast Asia, which is a pretty big gaming market as well. They essentially skipped PC gaming and went straight to smartphones.

For the wealthier East Asian countries like China, Japan, and Korea this isn't the case. It's more of a cultural thing due to various factors including

- Smartphones and mobile games became popular long before they became popular in the US. People are used to paying money for mobile games. It was this way from the very start.

- The Gacha model was popular in Japan even before mobile games. That's where the word comes from. Korea had similar machines.

- People take public transportation much more than in the US, which leads itself to playing games on the go (mobile).

- At least in Japan (less so in China and even less so in Korea), people prefer non-competitive gaming. That's why competitive games like DoTA, LoL, Counter Strike, etc, were never popular in Japan. Even Diablo with PVP never was popular.

loveparade··on Ask HN: Why do games (as media) make so much money?
Mobile game revenue makes up for more than 50% of all game revenue, and that number is only for the US. It's much larger in Asia, which is an even bigger gaming market.

For non-mobile games, most of the revenue still comes from mobile-game-like monetization schemes such as lootboxes/gacha/battle passes/skins/etc, not from the traditional sales.

These types of monetization schemes are incredibly lucrative, with many people spending hundreds, thousands, or tens of thousands of dollars over the lifetime of such games due to their addictive nature and gambling mechanics. Traditional media does not have an equivalent to these kind of monetization schemes.

It's not the size of the audience. It's the monetization scheme. The times when game revenue meant buying good old single player games without DLCs are long gone. Companies have long realized that's not where the money is.

loveparade··on Binance says banking partner will stop processing SWIFT transfers under $100K
1) Volatility is relative. USD Stablecoins are not volatile compared to the US dollar. It's the same. In other countries, the fiat currency is much more volatile than ETH or BTC. Like I mentioned above, 5+ years ago there was a push for retailers to adopt this stuff, but at the point stablecoins were not around, or unusable due to slow transactions or high fees. That's no longer the case.

2) True, I think the negative image and bad UX of retail crypto is a big reason of why retailers have not adopted it. I'm not sure this will continue forever though.

3) I would add that taxes are another reason retailers are vary of adopting crypto. Depending on the country, dealing with crypto taxes is a pain, and not worth the effort.

The magical solution? It's time, competition, and the free market. If crypto is truly cheaper and easier to deal with, startups will push it onto retailers to replace credit card incumbents. But these things don't happen overnight.

loveparade··on Binance says banking partner will stop processing SWIFT transfers under $100K
That's a matter of adoption. These things move very slowly. If you look at the adoption of debit cards, it looks similar. They were first released in the 1960s, and most retailers fought against them and said they would never accept them.

A few years ago there were valid technical reasons for why crypto is not being adopted by retailers in the form of slow transactions throughput and high fees. Nobody wants to wait a minute and spend $10 for their transaction to go through. These technical reasons no longer exist. L2s are extremely fast, secured by L1, and fees are small. And with the upcoming sharding upgrade Ethereum is going to be even more efficient. From a technical perspective, you could now accept USDT (or any other token) and have verification be about as fast as most card payments, with lower fees. But all of these are relatively recent developments (~1-2 years), and it will likely take a few years before all this technology trickles down into real-world adoption.

loveparade··on Ask HN: How has functional programming influenced your thinking?
The vast majority of challenges, bugs, complexity, and performance issues in software engineering can be reduced to managing state. State is hard. Immutable state solves a lot of these issues in theory, but in practice mutable state is the tradeoff you must make in many cases to get acceptable performance. The cost of this tradeoff is the increased complexity cost.

A good engineer understands the sweet spot of these tradeoffs. Use functional techniques with immutable data structures in the right places where performance is unlikely to become an issue.

loveparade··on Four thousand weeks
> So what's my point? The point is, whatever pursuits we undertake with the belief out there that something that I accomplish, accumulate or experience will bring me happiness and fulfillment is a futile endeavor.

I recently came across a term for this. Telic vs. atelic activities [0]. Telic activities are things with some terminal state, e.g. a typical goal-oriented project, or something like the act of getting married.

Atelic activities are those activities where the continuous process is the goal. Certain types of learning, being a good parent, and so on.

The important thing to realize is that in most cases it's not the activity itself that defines whether it's atelic or telic, it's how we approach it. You can make "traveling the world" a very goal-oriented activity with a checklist that you must get through, or you can approach it as a continous lifelong project where the enjoyment is in figuring it out as you go.

I believe this distinction is also important for work. Making something telic, e.g. with an agile process, comes with the danger of taking the day-to-day enjoyment out of an activity such as programming that you would normally enjoy. Of course there must be some amount of planning, but I think we've pushed too far into the goal-oriented direction that makes people miserable.

[0] https://en.wikipedia.org/wiki/Telicity

loveparade··on My PhD qualifying exam was a nightmare but I'm not letting it define me
I don't understand how "Black Lives Matter" and "BlackInTheIvory" is related to the content of this all. Why is this in there?

This reads like it was generated by ChatGPT with "Write a sad story about PhD qualifying exams but sprinkle in references to social movements about racial inequality"

loveparade··on Ask HN: Strategies for working with engineers that are too smart?
Have you tried explaining it to them in exactly the way you have written it here? What was their response?

In my experience, this happens quite often with people coming from academia because they're used to write throwaway code that doesn't get read by others and doesn't need to be maintained. It's only used temporarily to publish papers. However, most of the people I have worked with were aware of the fact that their code didn't follow software engineering best principles, they just didn't know how to do it better.

Another factor may be ownership. Sometimes people are excited to implement something because they like the idea of owning part of the stack, e.g. the caching implementation, instead of constantly implementing minor improvements in other people's code. Perhaps these people are looking for ownership of something more so than a specific complex solution?

loveparade··on Ask HN: What are some apps that feel like magic to you?
Google Maps. We're all taking it for granted, but imagine all the edge cases, complex real-world infrastructure and data analysis that goes into mapping the streets and buildings so well that it actually works, even in small towns.

It's much more impressive than Google Search. Given enough compute/storage, I feel like I can build a search engine. With Google Maps, I would have no idea where to even start.

loveparade··on “The current climate in AI has so many parallels to 2021 Web3”
> The difference for me is that Web3 was never shown to be at all useful.

Because nobody agrees what web3 even means, it's a totally useless term, not that dissimilar to AGI or consciousness, which nobody can properly define. People who hate crypto define web3 in terms of all the negative stuff, while people who have a vested interested in pumping token prices (like VCs) will define web3 in terms of all the potentially positive stuff.

I believe that the that fundamentals of crypto are incredibly useful and will eventually be adopted, and have been in meany places, but all the web3/NFT/etc stuff (whatever that means) is just useless hype and ponzi schemes.

Crypto has a much more negative image because it's full of scams, but that doesn't mean anything. It's just significantly easier to create scams with crypto because money is at the very base layer, than it is to create scams with AI. But that doesn't imply anything about the usefulness of these technologies.

loveparade··on Asking ChatGPT to write my security-sensitive code for me
I think it's just as likely to be the other way around. AI has been doing these hype cycles for for 30+ years now and every time it's "this is amazing and will take people's jobs" - But what always happens is not much change in the short term, and rather subtle changes in the long term. People underestimate how important the edge cases these models aren't good at are.

I also think ChatGPT is amazing, but historically it's not any more amazing that many of the previous AI-related developments for their time, none of which have quite lived up to the initial promise after the hype died. Just look at some of the robots from the 80s [0] that were expected to be in every household. And when chess was solved, of course we were just a few years away from AGI. In 20 years, people will look at ChatGPT and make fun of how cute it was, and we'll still be just a few years away from AGI.

People aren't blind, they are just realistic. Yes, current AI models will slowly be integrated into services and bring changes behind the scenes, but it's not going to be the big explosion of change a lot of people expect it to be.

[0] https://www.youtube.com/watch?v=jkOctWWsj-A

loveparade··on Ask HN: What are youre favorite languages to debug in?
I'll probably get a ton of pushback on this, but I never found huge value in debugging compared to using a strict type system and writing easily testable code. I haven't debugged, as in used an interactive debugger like gdb, in many many years, not counting occasions where I needed to reverse-engineer some C code.

I would prefer languages where the need for interactive debugging is minimized through the type system, good abstractions, and good testing frameworks. Functional languages (ideally strongly typed), or something like Rust, tend to fall into this category. And even for less strict languages, I find debugging to be significantly slower than spending upfront time writing testable abstractions, which usually means functional code that operates on immutable data structures when possible.

Debuggability is nice... but isn't not having to debug even better?

loveparade··on Less gym time, same results: Why ‘lowering’ weights is all you need to do
This article is a perfect example of

1. An academic publishes an extremely limited study, group size is 14 people that haven't exercised much, they do a single exercise, and they mention eccentric-only is particularly useful for rehabilitation. They make no outrageous claims whatsoever, they just put more numbers to something that is widely known and has been measured before.

2. Some popular news outlet wants to get clicks and runs away with it by making super general fitness advice out of it that's total nonsense. But you know, it's backed by Science (!) now.

loveparade··on Microsoft is preparing to add ChatGPT to Bing
It's easily verifiable, but it may still waste time. I've had many cases where ChatGPT makes up functions that do exactly what I need, but then I find out these functions don't actually exist. This may not happen very often for super popular languages like Python or Javascript where training data is huge, but it happens all the time for the long-tail of languages. In those cases, it would've been faster for me to do a regular search.

I do agree with the overall point though. If you understand when to use it and when it's more likely to give you nonsensical answers, it can save a huge amount of time. But when I ask it about a topic that I don't know enough about to immediately verify the answer myself I'm forced to double check the answers for validity, which kind of defeats the purpose.

The best queries to ChatGPT are cases where I know what the answer should look like, I just forgot the syntax or some details. Bash scripts or Kubernetes manifests are examples here, I know them, I just keep forgetting the keywords because I only touch them every few weeks.

And don't get me started about asking ChatGPT about more general topics in e.g. economics or finance. What you get is a well-written summary of popular news and reddit opinions, which is dangerous if it's presented as "the truth" - The big mistake here is that the training procedure assumes that the amount of data correlates with correctness, which isn't true for many topics that involve politics or similar kinds of incentives where people and news spread what conveniently benefits them and gets clicks.

loveparade··on My Youtube earnings
The most impressive thing to me is the initial period of 2-3 years, maybe more if the graph doesn't show the full history, with very little traction and negligible income. Of course that's expected, but sticking with it for so long without profits is not something most people can do, both emotionally or financially. Even if you love the topic you're making content on, the time and effort that goes into video editing isn't fun.
← PreviousPage 12 of 13Next →