antpls

In Zendesk, you get an Halloween pumpkin emoji instead of a thumb up. I always wondered if it was an annoying easter egg or a bug.

Some years ago, I evaluated Protonmail as a replacement for my personal gmail account.

When came the steps "can I easily move from this service?", I realized you have to _pay_ to export all your emails from the service. They make it super easy for you to open an account and receive emails, and then makes you pay if you want to get a copy of your own data.

I contacted the support to tell them it is likely illegal under European Data Privacy laws. They replied I can still export email for free one by one if I wanted to... (which is obviously not a valid answer when you have 5000 emails)

Then I looked in Swiss laws for a similar clause, and found that Swiss laws doesn't give users of online services the right to easily and freely get a copy of their data. It was a law proposal at the time of my research.

So yeah... Your data is so secure in Switzerland that you don't even own your data !

I was also worried about it, but then I learned about the USB-C to jack adapter cable. Look on your favourite shopping website.

That should buy us 2 or 3 more years of wired headphone, and let the wireless tech continue to improve.

Same, I have been using snap to run Firefox and Chromium for about more than a year (personal usage, not professional), and rarely had issues. I had one issue regarding a feature like WebGL or a file picker, but it was solved after 5 minutes of googling, and it was about running a single command line to give more permissions to the snap package. This way of distribution works quite well for big applications with many dependencies that you want to be automatically updated without fear of breaking some shared libraries.

I also feels more confident trying new applications with snap. I know I can easily install different versions and uninstall them without breaking the system.

Question regarding Fedora : I tried it in a docker environment. During some googling to troubleshoot issues, several results were found on the Red Hat commercial support website (so, not accessible). I'm afraid that a good part of advanced/niche documentations and knowledge base for troubleshooting is behind Red Hat commercial support. Is that the case in practice?

https://www.nhc.no/en/five-facts-about-the-referendum-in-bel... :

"The amendments to this article will make it sound as follows: The Republic of Belarus excludes military aggression from its territory against other states."

Either Lukashenko will tell Putin to remove Russian troops from Belarus, or Lukashenko is breaking the new constitution of his own country.

SIMD instructions as we know them today do not make a consensus among the actors and architectures. See : https://www.phoronix.com/scan.php?page=news_item&px=Linus-To... That makes them harder to implement in compilers and in turns to democratize SIMD.

There is an article on the web explaining the purpose of SVE/SVE2 ("Scalable Vector Extension"), which is supposed to be the successor of SIMD on ARM : https://levelup.gitconnected.com/armv9-what-is-the-big-deal-...

Extract : "[...] the addition of SIMD instructions has led to an explosion in the number of instructions, especially for x86. And of course not every x86 processor will support all these instructions. Only the newer ones will support AVX-512. The beauty of SVE is that the same code will work for both the super-computer and the cheap phone. That is not possible with the x86 SIMD instructions."

There is also a Java proposal to use SVE as the way of doing SIMD in the Java world : https://openjdk.java.net/jeps/417

The same principle will be extended on ARM to matrices with "Scalable Matrix Extension" : https://community.arm.com/arm-community-blogs/b/architecture...

We can speculate that everyone will migrate to ARM / RISC-V at some point, or x86 will have similar instructions.

No one mentioned video games settings?

It's a solved problem there, have profiles : "Low", "Medium", "High", "Ultra High".

They allow for a quick start, then as the user is more aware of the feature, the user can tune the predefined profiles themselves.

You can have independent profiles for each independent feature : input profiles (keyboard/gamepads), graphics profiles, audio profiles, etc

For example, on a Google account, that would be a single drop down list with those options: "Private and no personalized experience" (everything turned off), "Private with personalized experience", "Public"

Good reading ! It would be interesting to have other similar challenges, such as Euler, solved in idiomatic Tensorflow and Pytorch. Also some examples of more complicated state-of-the-art algorithms, such as sorting/graph/trees algorithms reimplemented in these frameworks.

It would be a great introduction to these frameworks for people who never touched anything ML-related, leaving the neural network content to later in the learning process.

Learning how to create differentiable algorithms and neural networks would be easier once the way those frameworks work is understood (ingesting data, iterating dataset, running, debugging, profiling, etc).

If you are starting with neural networks or differentiable programming, learning both the maths and the frameworks at the same time can be quite overwhelming

HN's Point of View doesn't represent the PoV of the 1 billion active iPhone owners.

Most people won't have any idea about the meaning of "hashes" and "databases". Not everyone is trying to actively fight the system and shit on everything, most people just want to live happily with their friends and family, they won't care that Apple scans their devices.

"Either there will be heads rolling at management, or Apple takes a permanent hit to consumer trust."

Oh god ! How wasn't all of this obvious to the top Apple management, but so obvious to epistasis! Damn, thanks man for correcting and leading Apple to the right track !

Tesla's Autonomous driving ads are banned in Germany since about a year now : https://www.reuters.com/article/us-tesla-autopilot-germany-i...

Not in Germany : Tesla' self-driving ads are banned in Germany since about a year now : https://www.reuters.com/article/us-tesla-autopilot-germany-i...

Since no one mentioned it in the comments so far, please add Hazelcast Jet to the long list of stream processors : https://github.com/hazelcast/hazelcast-jet

In order to understand the ecosystem, I ask myself this question : why did 4chan personally harass Audacity leader, but did nothing about the behavior of freechat leader?

To me, the downside of graph db is the non-standardized query language.

I tried Gremlin but it feels like an imperative DSL. Cypher queries are more readable, but are limited to Neo4J. I am looking forward for Open Cypher or maybe a variation of Facebook GraphQL.

One has to admit, Copilot raises many questions regarding global code quality, reviewing processes and copyright. It's a marketing success.

Another way to look at it : if an "AI" can predict what you would code next, it means your program is probably not that innovative, and was already created somewhere.

That doesn't look like a "leak", but more like the usual mass scraping of APIs.

An actual data leak from a breach would contain password hashes and private messages.

It means somehow, people can access that "leaked" data anyway, either with APIs or by paying LinkedIn

now, that happens in a zoom/slack I am not in, and awareness is much lower.

It sounds like all you need is a social graph of interactions from your chat software. That could lead to micromanagement deviance, but if the graph is open to all employees, I believe it would be fair and help everyone.

Great, but that's not a reason to force people who don't want to do it

The way I understand the comment :

With an interpreter, the JavaScript source is compiled to some bytecode, then the bytecode is executed in a loop, instruction by instruction.

With a compiler (not jit), the whole bytecode would be further fully compiled to some C apis before being executed, making it effectively an "ahead of time" compilation.

The catch is : since the code never was executed, there is very little static information to work with, so the compiled output is often just a giant unrolled interpret loop. The compiled version can be like 10% faster on very hot JavaScript loops compared to the interpreted version.

"In the Netherlands, the Dutch National Bank (DNB) now requires that crypto service providers looking to officially register with the central bank must demonstrate their compliance with verification requirements under the 1977 Sanctions Act. It involves establishing the identity and place of residence of the counterparty, screening it against the sanctions lists and establishing that this person or legal entity is actually the recipient or the sender."

I believe that is no longer true : "DNB has now revoked the registration requirement and said it will remove the wallet-verification measures as soon as possible."

https://www.coindesk.com/dutch-court-rules-bitonic-no-longer...

That's great, but it still performs 2 times worst than the best supervised model.

Also : "The discriminator itself is also a neural network. We train it by feeding it the output of the generator as well as showing it real text from various sources that were phonemized."

Is the "real text from various sources that were phonemized" a manually labelized database? If Yes, that step is supervised, which makes the whole thing actually supervised to some extent

Why is this flagged?

I like to believe to that theory as well.

But there is probably a simpler explanation : crypto is still experimental and Elon is pragmatic as new informations arrive. Where Bitcoin is the graal for some people, for Tesla it was only another real scale experiment about a payment method.

Could be alleviated by creating yet another permission at the browser level : "allow to link to local applications"

The reactions on Twitter are crazy, the Bitcoin fanatics who never cared about other cryptocurrencies are loudy.

But that's a great move for crypto as a whole : once the environmental issue is resolved in people's mind, crypto (not necessarily Bitcoins) might attract many more investors, and we will finally have GPUs!

No mention of https://frictionlessdata.io/ dataset metadata format which is also used by Kaggle

Thanks :

1. In distributed computing, there are usually minimal communication because nodes communication is costly, the bottleneck can still be the hot loop on each nodes.

2. When optimizing the hot loop only, it can be quickly written in Cython or C, since the scope is usually a lot smaller, so no need to use Go.

People need to focus on solving business problem first. "memory layout" and "cache usage" should be abstracted with premade data structures and algorithms into a library. You could then use "CacheFriendlyHashMap" class as a drop in. Does such library already exist for any language ?