Oh, a jump in English speaker? 👀

Awesome, thank you Sebastian 🙏

Few. Many might forget this 😅

NDK or nostr-tools will probably do

Thank you. Yes, i just realized it now 😅

It changes with this commit

https://github.com/hoytech/strfry/commit/7da862eb913b3cffffbd003ba8c408999b4dcbf5

Nice. Is this for strfry write policy plugin, Takumi-san?

Interesting question. I also wonder about that. 🤔

I think Tutanota/Tuta article might answer our question a bit

https://tuta.com/blog/images/NSA_Nato_alliance.webp

"Big brothers are watching you"

Even the so called 'neutral' countries like Switzerland can't be called as really 'neutral'

"Although Iceland or Switzerland claim to be outside the reach of the Fourteen Eyes countries, both of these countries’ law enforcement agencies cooperate regularly with Europol requests which makes their services just as prone to turning over user information as any service based inside of the Fourteen Eyes." (Tutanota/Tuta article)

https://tuta.com/blog/posts/fourteen-eyes-countries

>Anser "no" to request from other countries

It is hard to avoid most world (assume most world/other countries definition that you have said are related to NSA/NATO/West Alliance) influences unless we want to store our data in Russia/China/BRICS alliance countries. There is no real 'neutral' country who can stand against both current geopolitical alliance in the worst scenario of they forces (military and politically) other countries to join one of both sides.

Possible Solutions:

- Relay operators might probably play on both sides by hosting relays in NSA/NATO/West Alliance and BRICS countries altogether (example: First mirror relay in Russia, Second mirror relay in Germany). At least if the content stored 'pis*ing' one of the alliance groups then it will probably still fine hosted in another group.

- Relay operators also need good OPSEC and serve their unknown Third relay with TOR onion hidden services (onion domain).

Safe? Protected in bunker with high grade security in non 9-eyes country? 😅🤔

Thank you. Hopefully this will cover most connection bug case. 😅

Becak (can carry 2 or 3 people) almost no carbon emitted 😅

Jokes aside. I think maybe it can be calculated at least based on efficiency in several values consist of:

- How far it can reach the distance

- How many people that it can cover during one trip (capacity)

- How many energy spent in one trip

Don't know who have done the research 😅

It will include your follows + their follows nostr events (notes/posts, likes, zaps, etc) when we used it instead of all events from global. filter.nostr.wine can be useful for global feed relay by limiting the events only to our follows + our follows of follows (their follows) using web of trust. I think you can read more on filter.nostr.wine for better explanation

Thank you Kirino-san 🙏

Yes, nostr bring us the possibility in "freedom of speech" and also "freedom to see/hear" what we want. My project is mainly focused on assisting "freedom to see/hear" for users by giving them configurable options/tools on relay.

Thank you 🙂

Yes major big relays probably have strict rate-limit rules. The solutions are directly add them to our write relay list and not fully depend on blastr/broadcaster relay. Maybe, 5 minutes later we can try broadcast again including blastr/broadcaster relay. 😅

Well, the problem of blastr and broadcaster relay like filter.nostr.wine that you have mentioned in your other posts are they can probably be rate-limited by the target relays because target relays see them as spammy connections (unless they have multiple IP) 😅

Maybe they were rate-limited when your notes were posted. I have seen blastr has some 'retry mechanism' in their code but i don't know with how filter.nostr.wine do behind the scene

I also still add my main relays but set them to mostly write only. Relay like damus and nos.lol usually can't get events from blastr 😅

Yes, at least on my experience. I blast some of my notes through mutiny blastr. 5 minutes later, i have checked and found that they were broadcasted to 100+ relays based on https://nadar.tigerville.no reports

There is also possibility Amethyst has some problems when you first post it. Maybe you can try close Amethyst after you posted it and open again and press "Broadcast" on the same post later

Mutiny blastr? Usually it works fine 🤔

ChatGPT-3.5 MMLU score report:

https://chat.lmsys.org/?leaderboard

Based on report from Open LLM Leaderboard benchmark results in Huggingface, Yi-34B model achieve the best in all score averaged (averaged across multiple metrics) for open model.

https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard

Also, it has better performance compared with ChatGPT-3.5 based on MMLU score (ChatGPT-3.5 = 73.4).

Huggingface is one of the best place to catch up with AI/ML research or applications 🙂

Thank you, it still needs many improvements to achieve that kind of feat. Hopefully even more better softwares will come 🙂