< Back to the archive

Like what you see? Subscribe here and get it every week in your inbox!

Issue #224 - June 25, 2023

Here are the top threads of the week, happy reading!

Top comment by kunwon1

I have worked professionally with access control and surveillance. I can give you two manufacturer recommendations: Axis and Geovision.

Axis cameras are high end and expensive, but they will, in my experience, do anything an IP camera could reasonably be expected to do, and they will do it well. They are European in origin and are available from various retail outlets to ship this week.

Geovision cameras are low end and not expensive. They are Taiwanese in origin and are pretty easy to find.

I have personally configured a wide range of cameras from both of these manufacturers and I have never needed an app or internet connectivity. It's been a few years since I looked at Geovision's product lineup though, my information is not 100% current. I don't have any specific camera recommendations. If I were setting up a home NVR today, I would buy Geovision cameras and put them on an isolated network.

Both of these manufacturers are nominally ONVIF compliant (ONVIF compliance is a mixed bag and can't be fully trusted from any manufacturer IMO) and have readily accessible RTSP streams

Top comment by rubinlinux

Reddit was a co-op between a few groups of people.

General low effort content scrollers

Power users and mods who appreciate creating and contributing to make a community

Advertising interests

The deal was, the ad seeking was for the mainstream strollers, and the contributing power users and mods could opt out of the bs. The community builders get a nice environment for their community, the scrollers get content, and the ad people get to shiw thejir ads.

I always thought spez understood this. Its why the api existed. Its why old.reddit.com existed. It was the commercial machine's compromise to the content generators in exchange for the moderating and commenting.

But he seems to have forgot. I wonder why?

Without the compromise the whole thing falls apart. Reddit becomes digg.

Top comment by marginalia_nu

A big drawback, and what killed many of the original forums, is spam management and maintenance. It's a very lucrative target to attack for two reasons. Popular public forums tend to be very central in pagerank, so if you manage to post comment spam, that's a big win. Forums also act as large databases of usernames and (maybe) hashed passwords, so if you manage to find a vulnerability in the software, you can sell those for decent cash.

This means that forums are always under attack on multiple fronts and require time consuming moderation and maintenance around the clock.

Centralized solutions like Discord and Reddit can use economies of scale to tackle the problem in a much more cost-effective way.

Top comment by eatonphil

Read good books like Designing Data Intensive Applications and High Performance Browser Networking.

Learn how to communicate with folks outside of engineering. i.e. by spending time with peers in product management, customer support, marketing, etc.

Do projects that push your understanding: https://austinhenley.com/blog/challengingprojects.html and https://austinhenley.com/blog/morechallengingprojects.html.

Join a startup if you haven't worked for a startup. Join a massive corporation if you have only worked for startups.

Read more source code of projects and your dependencies. Make contributions to the docs, improve the error messages of projects you use.

Write blog posts or just notes to yourself to organize your thoughts and ensure you understand what you think you do.

Join companies or teams within your company where you feel challenged to grow, not being complacent. Don't worry about feeling like an idiot.

Just a couple of ideas.

Top comment by jerf

Cookies was a simple concept that politicians could glom on to and legislate around, but, yes, they're quite unimportant in the grand scale of things.

In the end, you've got two things to work with: Things you can convince the browser to actively identify itself with, and the things you can track regardless.

Cookies are in the first category, but they are not alone. You can get things as simple as presenting an entire site with customized URLs that track a user through querystrings being appended to everything with an identifier. You can track certain caching differences. You can program a website to use local storage and submit a token on every URL click with a fairly simple handler. This isn't even remotely a complete list.

In the second category, you've got IP address, browser versions, various settings... see something like https://www.amiunique.org/ .

In a nutshell, your rich browser experience leaks so much data along so many axes that it is essentially inconceivable that you could ever prevent yourself from being fingerprinted. What you can do is try to detach that fingerprint from a real person, to a certain extent rotate what you can, etc. But in reality you can't be shipping up kilobytes of header information on each web request and expect there isn't something in there that can track you.

https://www.amiunique.org/fingerprint says I'm 100% unique; with all the red lighting up I'm not surprised.

Top comment by __rito__

- Have crystal clear Mathematical foundations, as in why this formula/method the way it is, rather than being able to solve college/HS test problems. Really solid footing in Differential Calculus and Linear Algebra is necessary.

- Know the Statistical language that you learn from a basic college-level Stat 101 course. Be able to translate normal sentences into those using Statistical notation, and be able to read easily. Also, know basic Statistics.

- You already know programming, I assume. Learn Python if you don't know already. It's really easy.

- There are a number of paths you can go from there. Here's what I did.

-- IBM Data Science Professional Certificate (not deep at all, but lays out the landscape well; did it in a week)

-- Machine Learning for Absolute Beginners by Oliver Theobald which you can finish in an evening.

-- Machine Learning Specialization by Andrew Ng on Coursera.

-- Deep Learning Specialization by Andrew Ng on Coursera.

-- fast.ai course.

- Learn PyTorch really well. I suggest Sebastian Raschka's book.

Now from here, you can chart your own path. You can choose NLProc, Vision, RL, or something else.

I went towards Vision. And I do Edge AI as hobby.

I was in the last year of college as a Physics undergrad, when I was hired to do Vision modelling/research for a non-flashy company in 2021. Finishing my CS Master's next month and starting to look for PhD. I worked in the same company for the ~2.5 years.

EDIT: If you want a job in big tech, grind Leetcode, and learn about system design, study Machine Learning systems, and be able to design them. Chip Huyen has a good book as I hear. 6-7 rounds of interview is common in Meta/Google. DL hackathon awards, open source contributions are significantly helpful.

Top comment by fuzzythinker

Ya, it's been said that "it is part of the issues to be fixed in the next UI update" for like 10 years now.

To me, the most annoying part of using HN w/o other clients on mobile is not this, which can be addressed with extra care or zoom in. It's the idiotic decision to somehow think the post text is less important than comments and so the font color must be set to be lower. It's so low that it's unreadable in mobile. For long posts like many show-HNs, I have to just skip reading it until I get back to desktop, which isn't only not ideal, but also defeats the purpose of show-HNs since the ones that doesn't stay at top are the ones needed all the attention and HN is not helping by making it unreadable.

Ya, a simple css color change to match comments' color need to be part of the fixes to be addressed when there's a blue moon.

Top comment by kristianp

Similar topics in the past year:

Ask HN: What is the most mind expanding book(s) you have read till date? 85 comments, 6 months ago https://news.ycombinator.com/item?id=34193766

Ask HN: Which book would you pick to re-read for the rest of your life? https://news.ycombinator.com/item?id=31968169

Ask HN: What book have you re-read 3x or more? https://news.ycombinator.com/item?id=32712496

Ask HN: What book(s) had a profound impact on your life and why? https://news.ycombinator.com/item?id=34314364

Ask HN: Best books read in 2022? 249 comments https://news.ycombinator.com/item?id=33849267

Ask HN: What is the best thing you read in 2022? 259 comments https://news.ycombinator.com/item?id=34055123

Ask HN: Name 3-5 books that had the most impact on your career and knowledge? https://news.ycombinator.com/item?id=32896299

Top comment by fy20

- Culture and language. Even if you compare US vs UK, the UK is much more European in it's approach to work. If you want to exploit people and have them dedicate their life to your company, better stick to US workers.

- Working hours. If you are on the West Coast, good luck finding someone in Europe who will be available at 3pm your time.

- Superiority complex. I get the impression a lot of Americans treat everywhere else like it is a third world country. I work for a US company that has a branch in a post-Soviet EU country, and when the C-levels were discussing expanding our team, they were talking as if there would be a queue of people waiting at the door to interview.

Taxes and employment laws are really not an issue, there are plenty of ways around it, especially if you are a small startup. As others have pointed out, the simplest way is to treat someone as a 1099 independent contractor. You pay them gross, and let them deal with all their own tax responsibilities. If you want proper employees there are agencies everywhere who will deal with this for you, and in a lot of cases government agencies will help you and / or give you tax breaks as they want you to invest in their country.

(Source: As a European, I've worked half my career remotely for US companies)

Top comment by arp242

I have come to believe that by and large "best practices" are actively harmful as they're mostly pounded by inexperienced programmers who lack the ability to judge the applicability or zealots who believe everything should be done the One True Way™, and they (ab)use these kind of "best practices" as an argument from authority. Everyone else just says "X is better because Y", which is an actual argument that can be discussed (engaging with actual arguments discovers what the best solution for the particular situation is).

Most "best practices" are just someone's opinion anyway, and little more. The exception to this are some things like "don't use mktemp(), use mkstemp()", but these are usually not called "best practices" but "footguns" or "things that will get you in to trouble sooner or later".