HNHacker News
TopNewBestAskShowJobs

natdempk

2,915 karma · joined May 31, 2013

https://dempkow.ski (& https://natdempk.io)

x: @natdempk

natdempk at gmail dot com

submissionscomments
natdempk··on You Can't "Win" in Tech Unless You Cheat
> See the pattern? The biggest tech successes of our era didn’t win through superior technology or innovation.

- PayPal won by creating technology that makes it easier / possible than any alternative to send and receive money via the internet.

- Uber won through creating an app that lets you hail a ride and creating a marketplace for that.

- Airbnb won through innovating on the social idea of paying someone and creating a marketplace for that.

- Facebook won through creating an innovative social networking technology and creating an ad marketplace on top of that.

- Amazon won through creating an innovative website and marketplace for goods that offered innovations to the consumer like quick shipping times and risk-free purchases via easy returns.

- Tesla won through literally creating a product (EV) and charging network for EVs that people are willing to pay for.

This article is arguing in bad faith. All of these services won because they created something that solved a problem and created value for millions or billions of people, either making plenty of money along the way, or with high potential to make money at any given point in time. If this is cheating, then all of capitalism and the innovations that scale and improve life for the world via capitalism are all "cheating". Build something people want and whatnot.

natdempk··on Everyone knows your location: tracking myself down through in-app ads
Yeah my mind also immediately jumped to fingerprinting. Somewhat required for anti-fraud to some extent, but also obviously used for more than that.
natdempk··on OpenAI says it has evidence DeepSeek used its model to train competitor
I don’t think anyone is really suggesting they stole COT or that it is leaked, but rather that the final o1 outputs were used to train the base model and reasoning components more easily.
natdempk··on Supreme Court upholds TikTok ban, but Trump might offer lifeline
> It seems all the folks who “have it on good authority” that this is a dangerous propaganda tool, can’t share what “it” is.

No need to speculate too hard here, there are plenty of examples of censorship on TikTok: https://en.wikipedia.org/wiki/Censorship_by_TikTok

Censorship is a form of propaganda, and even the very obvious/reported examples we've seen reported over the years are pretty bad. And you have to assume that there is more going on than is actually reported/noticed, especially in subtler ways. It's also just obvious it's happening in the sense that the Chinese government has ultimate control over TikTok.

natdempk··on JavaScript Benchmarking Is a Mess
Does anyone know how well the JIT/cache on the browser works eg. how useful it is to profile JIT'd vs non-JIT'd and what those different scenarios might represent in practice? For example is it just JIT-ing as the page loads/executes, or are there cached functions that persist across page loads, etc?
natdempk··on Scientists are learning why ultra-processed foods are bad
The Slime Mold Time Mold article is maybe an interesting starting point, but really not great analysis... see this thorough rebuttal around the Lithium stuff that has not been responded to: https://www.lesswrong.com/posts/7iAABhWpcGeP5e6SB/it-s-proba...

TBH a lot of the discourse around this annoys me because it ignores sort of fundamental obvious stuff like food/snacks got way tastier, access to unhealthier food is higher, we live a way more sedentary lifestyle, takeout got more affordable, studies show that people overall consume more calories from unhealthier sources, you can go to the grocery store and see you should basically avoid much of the food there, food spikes your glucose higher, more artificial ingredients, switch from fat to sugar, more processed ingredients, etc. and yet everyone says "well it can't just be that" because XYZ specific study of some quality and timeframe took one of those axis in some degree of isolation and showed that _maybe_ it's not that specific thing only. Kind of a missing the forest for the common sense trees aspect. Obviously we don't understand the whole picture of obesity, but there's a degree of denial around common sense health stuff that's really weird in the kind of "rationalists try to figure out obesity" writing. I really think we should work more from a place of "you don't see people who do moderately intense exercise regularly and eat healthily being overweight/obese" as a baseline...

natdempk··on History of Hangul
Anyone know if there is a particularly great app/website out there for learning Korean? Ideally opinionated, low/no "figure out what to learn yourself", Anki + AI-powered for maximum gain/seamless review/ease of getting more reading + writing variety, and easy to use on-the-go/on-mobile.
natdempk··on Vampire Number
Cool, thanks!
natdempk··on Vampire Number
But why is it called a Vampire Number in the first place?
natdempk··on Internet Archive: Security breach alert
The current community notes I see on these tweets just basically say: the Internet Archive is not part of the US government.
natdempk··on Distributed transactions in Go: Read before you try
Well, the problem you run into is that you kind of want different datastores for different use-cases. For example search vs. specific page loads, and you want to try and make both of those consistent, but you don't have a single DB that can serve both use-cases (often times primary DB + ElasticSearch for example). If you don't keep them consistent, you have user-facing bugs where a user can update a record but not search for it immediately, or if you try to load everything from ES to provide consistent views to a user, then updates can disappear on refresh. Or if you try to write to both SQL + ES in an API request, they can desync on failure writing to one or the other. The problem is even less the complexity of keeping the index up to date in realtime, and more that the ES index isn't even consistent with the primary DB, and to a user they are just different parts of your app that kinda seem a little broken in subtle ways inconsistently. It would be great to be able to have everything present a consistent view to users, that updates together on-write.
natdempk··on Distributed transactions in Go: Read before you try
You have a distributed system on your hands at that point so you need idempotent processes + reconciliation/eventual consistency. Basically thinking a lot about failure, resyncing data/state, patterns like transactional outboxes, durable queues, two-phase commit, etc etc. It just quickly gets into the specifics of your task/system/APIs so hard to give general advice. Most apps do not solve these problems super well for a long time unless they are in critical places like billing, and even then it might just mean weird repair jobs, manual APIs to resync stuff, audits, etc. Usually an event-bus/queue or related DB table for the idempotent work + some async process validating that table can go a long way though.
natdempk··on Distributed transactions in Go: Read before you try
Pretty great advice!

I think the one thing you can run into that is hard is once you want to support different datasets that fall outside the scope of a transaction (think events/search/derived-data, anything that needs to read/write to a system that is not your primary transactional DB) you probably do want some sort of event bus/queue type thing to get eventual consistency across all the things. Otherwise you just end up in impossible situations when you try to manage things like doing a DB write + ES document update. Something has to fail and then your state is desynced across datastores and you're in velocity/bug hell. The other side of this though is once you introduce the event bus and transactional-outbox or whatever, you then have a problem of writes/updates happening and not being reflected immediately. I think the best things that solve this problem are stuff like Meta's TAO that combines these concepts, but no idea what is available to the mere mortals/startups to best solve these types of problems. Would love to know if anyone has killer recommendations here.

natdempk··on How to succeed in MrBeast production (Leaked PDF)
You mean he talks more like a normal person talking about a product rather than a Youtuber going over the top with everything? The fact that he is genuine is a big part of his appeal.
natdempk··on Ask HN: Why Is Stack Overflow Fading Away?
Good to know, for some reason my sort sets to `Date modified (newest first)` which seems bad also and I'm certain I never set it this way intentionally, but I guess that explains the issue and I was wrong that they have entirely let this problem languish. Thanks.
natdempk··on Ask HN: Why Is Stack Overflow Fading Away?
This is basically a product failure... I understand the distinction, and yet a negatively voted accepted answer shows up first above a highly-voted non-accepted answer. What is the point of the accepted answer in this scenario? Why rank it first? etc.
natdempk··on Ask HN: Why Is Stack Overflow Fading Away?
Yeah, the moderation is truly broken, along with the "accepted answer" framework, and in general the approach to knowledge curation. Kind of amazing to me that the founders/staff of the site didn't try to turn this around.

To add an anecdote: The last question I bothered to answer was one where the accepted answer was a very-specific fix, and a more generic fix was in a comment below that which was better and more directly addressed the root problem and would work for any user encountering an exception (accepted answer was workable, but working at the wrong layer of abstraction to actually solve the problem). I pulled that out into its own answer. Looking back now at that question, the poor "accepted answer" which won't work for anyone hitting this error because it references a specific class in the user's question which won't exist for any other users is still accepted with -5 and the better answer is below the fold at +16. This is pretty typical across a lot of questions. The fact that SO doesn't automatically handle this case is basically a failure of the site's abstraction model and algorithms over answers.

For a site where the long-term value is ostensibly curating high-quality answers to the maximal number of questions, the best answers languish, and the questions and answers don't get sufficiently refined/updated over time. Arguably you'd want something closer to a wikipedia article about each problem that gets built out and updated over time if you want to provide canonical information about problems. Similarly I think the idea of closing things that are close, but different as duplicates has failed. These are often sufficiently different that the details are interesting and probably would provide value/activity/detail to the site. There should probably be some way to roll these up into a higher level article/topic to cover variants of problems, related cases, etc. This could start to act as pillars or knowledge-hubs within SO to get to a place of more canonical information or a more "tacit/practical wikipedia". Really not sure why things stopped at Q&A and seemed to stagnate where they did.

Though, they seemingly achieved profitability and sold for $1.8B without doing any of this, so what do I know. :) Probably the right move was focusing on other things like launching new communities, and making money for the company.

natdempk··on DOJ sues realpage for algorithmic pricing scheme that harms renters
Similarly companies like Pave also do data collection and sharing to create “benchmarks” around salary and report out salary ranges for comparable companies and locations. I was wondering where the line is drawn between something like this and Realpage? Or even if you do some detailed survey about compensation and distribute it, at what point does this become anticompetitive? All of this does seem to rapidly be becoming the norm with regards to how companies set compensation levels. Curious if anyone with more expertise knows the answer.
natdempk··on We don't know how bad most things are nor precisely how they're bad
Well, it's important to differentiate what actually matters in cases like this otherwise we just get into a weird luddite place of complaining about things that don't actually matter.

Here it is the listening medium and environment where an orchestra hall is a physical building built at a cost many orders of magnitude more expensive than speakers specifically for the experience of listening to live music compared to listening to something on a home speaker or phone.

To me It's interesting that you assume Spotify is the problem, whereas its actually the listening medium you are using Spotify through. In the case of audio, you could listen to Spotify recordings on sound systems that will basically blow you away in the same way a live orchestra will, but they cost probably 10's of thousands of dollars. For example at the higher-end of setups people describe it being hard to differentiate between concert recordings and actual concerts, and this should be expected with what we know about audio.

I do agree though that quality degradation as a means of meeting cost and false-repackaging of that as higher quality is annoying. Especially in streaming video, things are advertised as 4K, but that ignores bitrate and we have the potential infrastructure to deliver experiences that are way better like streaming the equivalent bitrate of a Bluray, but we've kind of fallen into a place of "good enough" where most people don't know or appreciate what is possible. Audio is kind of similar, but there is maybe more of an argument that cost of listening systems means we're at a happy-medium of cost vs. quality and you can pretty much buy the level of experience you want there.

natdempk··on We don't know how bad most things are nor precisely how they're bad
> Uncompressed audio is radically different and richer than low fi digital compression, but almost everything we hear everywhere is coming through streaming services.

Out of curiosity, have you blind A/B tested yourself with the various streaming service "high" quality MP3 encodings vs. lossless? (like https://abx.digitalfeed.net/) I agree that there is some difference that you can hear in certain parts of songs listening on high-end equipment/headphones (stuff that is greater quality than >99.9% of the listening population has), but really to me it would be hard to call it radically different. Obviously if you are below 320 MP3s there are more noticeable compromises that do ruin things one you notice.

natdempk··on Ask HN: Have you noticed your songs being swapped out with inferior versions?
You gotta provide platform + examples for something like this.
natdempk··on LG and Samsung are making TV screens disappear
If you are unhappy with Plex on Apple TV, check out Infuse. You have a pay a little yearly for it, but it does a much better job of supporting higher-end/advanced media features and the app is nicer than Plex. It can seamlessly connect to a Plex server as well.
natdempk··on Initial details about why CrowdStrike's CSAgent.sys crashed
Most enterprises these days also run stuff like Crowdstrike (or literally Crowdstrike) on their macOS deployments. Similarly Windows these days is bundled with OS-level antivirus which is sufficient for non-enterprise users.

Not in the security industry, but my take is that basically the desktop OS permissions and security model is wrong for a lot of these devices, but there is no alternative that is suitable or that companies are willing to invest in. Probably many of the highest-profile affected machines (airport terminals, signage, medical systems, etc.) should just resemble a phone/iPad/Chromebook in terms of security/trust, but for historical/cost/practical reasons are Windows PCs with Crowdstrike.

natdempk··on Google begs court for relief from Epic Games' Play Store demands
I would encourage you to read more about the two cases and the differences between them and the rulings. Stratechery has some good articles though some are paywalled.
natdempk··on Biden signs TikTok bill into law, starting clock for ByteDance to divest
It's a rebrand + censorship then? I'm not sure the distinction you're trying to make exactly, it's not like Douyin is a completely different thing from Tiktok, so I think "completely separate platforms" is also a stretch. The core technology and parent company/ownership is the same for both.
natdempk··on 15 Note Poly Tempo Pendulum [video]
Ha, coming full circle :) that guy has lots of cool stuff.
natdempk··on 15 Note Poly Tempo Pendulum [video]
If you want to go even further: https://www.youtube.com/watch?v=SthcxWPXG_E
natdempk··on What I think about when I edit (2019)
When I learned about passive voice, scientific writing was often mentioned as an explicit exception. What you say makes sense when you are abstractly describing something that happened, but also I'd argue that "We added 30 grams of goop to the beaker" or "Added 30 grams of goop to the beaker" is clearer than "30 grams of goop was added to the beaker". Depends on the context and style conventions of course, but often passive voice is still poor/indirect writing in many scientific contexts.
natdempk··on How I keep myself alive using Golang
Actually I did read the whole thing before posting... There are different CGM models where some require scanning and some report continuously. My reading was that the author's model seemed to only support scanning thus the need for the hat, rather than periodic updates to an app that newer models provide, but it doesn't seem 100% definitive from the article as I don't know if on newer models data export might require scanning? The point of my comment was that rather than all this work, getting a different device could more-or-less enable the desired functionality without so many workarounds and the bulky hat device by eliminating the whole scanning/inconsistency aspect. Though if you want to pull the raw datapoints etc. and export them elsewhere, then there are still more complications. The non-scanning ones might still be a bit unreliable as well, not sure where that aspect is coming from.
natdempk··on How I keep myself alive using Golang
Not sure exactly what you mean by "getting data" but recent versions of both Libre and Dexcom have apps where you can view graphs of data on your phone with no tapping step.
← PreviousPage 2 of 13Next →