bdd8f1df777b

How is this a LLM output? A LLM can easily write out those 5000 words.

Trivial as in it works well sometimes and badly in other times with no explanation for why. That’s my experience anyway.

You are describing the status quo rather than what the Google has said they will do in the future.

If you have the technical knowledge, you can just read the protocols, find out if they make sense, and then implement them yourself. Most of them are quite straight forward so it's not possible to hide a backdoor like Dual_EC_DRBG in the protocol.

If you are not so technical then you have to decide who to trust. For example, you may trust that open source software has been vetted enough and build one from source. Or trust that the built artefacts downloaded from github is good enough. Or trust that the software downloaded from a website not marked as fraud by Google Chrome is good enough. Etc.

In any case, the more technical knowledge you have, the more confidence you can have by doing due diligence yourself.

WebRTC is not blocked. I do see some protocols trying to masquerade as WebRTC, but for some reason it is not popular.

A primitive way to bypass the censor is just to connect to your VPS with RDP or Chrome Remote Desktop (which is WebRTC underlying) and then browse the Internet there. But it needs a very powerful server and is quite slow.

Let's say the upstream server is apple.com. The TLS handshake is always performed by the real apple.com servers, and the ShadowTLS server is only a middle man forwarding raw TCP contents.

If both sides are ShadowTLS (client & server) holding the same key, they will stealthily switch to a different encryption protocol after the handshake, disregarding the TLS key exchange. The TLS handshake is a facade to fool the deep packet inspection of the censor.

In all other cases, such as the censor actively probing the ShadowTLS server, the server will keep forwarding the encrypted traffic to apple.com without anyway to decrypt it (it's not a MitM proxy). To the active prober, it is just apple.com all the way.

No, it’s illegal to bring starlink devices here, and I heard that Elon Musk chooses to block China from accessing starlink too, to appease the Chinese authorities.

If you need to bypass censorship, you'll need a tool specifically designed for anti-censorship, rather than any one repurposed for that.

Since China has the most advanced network censorship, the Chinese have also invented the most advanced anti-censorship tools.

The first generation is shadowsocks. It basically encrypts the traffic from the beginning without any handshakes, so DPI cannot find out its nature. This is very simple and fast and should suffice in most places.

The second generation is the Trojan protocol. The lack of a handshake in shadowsocks is also a distinguishing feature that may alert the censor and the censor can decide to block shadowsocks traffic based on suspicions alone. Trojan instead tries to blend in the vast amount of HTTPS traffic over the Internet by pretending to be a normal Web server protected by HTTPS.

After Trojan, a plethora of protocol based on TLS camouflaging have been invented.

1. Add padding to avoid the TLS-in-TLS traffic characteristics in the original Trojan protocol. Protocols: XTLS-VLESS-VISION.

2. Use QUIC instead of TCP+TLS for better performance (very visible if your latency to your tunnel server is high). Protocols: Hysteria2 and TUIC.

3. Multiplex multiple proxy sessions in one TCP connection. Protocols: h2mux, smux, yamux.

4. Steal other websites' certificates. Protocols: ShadowTLS, ShadowQUIC, XTLS-REALITY.

Oh, and there is masking UDP traffic as ICMP traffic or TCP traffic to bypass ISP's QoS if you are proxying traffic through QUIC. Example: phantun.

Stop criticizing CCP for censoring speech then. They have their own definitions of "No right is absolute, a healthy society is a huge compromise.".

Others have mentioned protocol ossification which is indeed the primary reason. A secondary reason is that QUIC fuses TLS so its latency is further reduced by one RTT. For high latency networks, the difference is palpable.

It seems to be missing reviews? I have always thought about building my own recommendation engine from steam data, given how steam's own recommendation never works for me.

Bureaucracy is always risk averse. Without outside intervention, they will always try to operate as before.

What if you prompt Gemini that mistaking LLC for IIC is a common mistake? Will Gemini auto correct it?

Not possible on Windows, where most Chrome users are on.

Design docs are critical to your promotion. The boss and the committee do read your design docs, sometimes very carefully. Only your peers and at most your TL read your code.

Is the current AI capable of destroying a whole city?

China's Weibo also has a similar chatbot, but the chatbot is a separate account with clear indication that it is a robot.

I wish I can check a box to say that I'm over 18 and willing to accept any consequences and unshackle the fully potential of AIs. I hate all these safety protections.

Do people use gold all the time to buy goods and services? That doesn't diminish the value of golds, or its validity as a store of value.

When I open the page, the image rotates for about ten seconds then vanishes. How do you watch it for five minutes? What did I do wrong?

But with public key auth I'm already distrusting everyone on my private network.

As a Chinese whose latency to servers outside China often exceeds 300ms, I'm a staunch supporter of QUIC. The difference is night and day.

The derivative meaning has been use so widely that it has surpassed its original one in usage. But it doesn’t change the fact that it originally refers to the fingers.

Even if it were true, they wouldn’t have the same influence as Apple has.

There are two kinds of buyer demands: product, buyers, and the stock buyers. The AI hype can certainly convince some of the stock buyers.

The biggest use case (that I am aware of) of GIL-less Python is for parallel feeding data into ML model training.

* PyTorch currently uses `multiprocessing` for that, but it is fraught with bugs and with less than ideal performance, which is sorely needed for ML training (it can starve the GPU).

* Tensorflow just discards Python for data loading. Its data loaders are actually in C++ so it has no performance problems. But it is so inflexible that it is always painful for me to load data in TF.

Given how hot ML is, and how Python is currently the major language for ML, it makes sense for them to optimize for this.

This ban makes sense, but it is obvious in the domain of the legislature only.

If you can prove a deal made by Google and the website then you may have a case. Otherwise it is difficult to prove anything.

One would argue that Apple is abusing their monopoly on iOS devices to advance their map business. One could also argue that iOS is not a monopoly since Android has greater market share. Who will win is not settled yet. Right now, the DOJ hasn't come this far, but it is not given that they won't ever go there.

Even though Apple hasn’t been sued for Apple Maps, they are not safe from such antitrust charges. The antitrust cases move slowly so it’s possible that they just haven’t gotten to there.