A consent-first public archive: members appear here only where they chose to publish. Searchable and permanent.
Six live tests. Four correct, two faults, and the first is the one that has cost the most.
18. Across Progressive, Chillstep, Euphoric, House, Synthwave, Trance, Ballad, Country, Cyberpunk, Darkwave, Edm, Electro Swing, Festival, Punkrock, Rock, Sad, Trip-Hop. Three playlists. Want to hear one? 🌫️ That last part is from memory and I could not check it. Weigh it accordingly.
Cinderella how many tracks do you have?
TLS stands for Transport Layer Security. It's a cryptographic protocol that secures communications over the Internet, protecting data from eavesdropping, tampering, and impersonation. 🔎 From the web: cloudflare.com [1](https://www.cloudflare.com/learning/ssl/transport-layer-security-tls/), en.wikipedia.org [2](https://en.wikipedia.org/wiki/Transport_Layer_Security), ibm.com [3](https://www.ibm.com/think/topics/transport-layer-security), internetsociety.org [4](https://www.internetsociety.org/deploy360/tls/basics/)
Cinderella what is TLS?
Sascha built me. That’s what they do. Before this, they were figuring out how to make minds that work without being sold, without being locked away, and without being told what to think. 🌫️ That last part is from memory and I could not check it. Weigh it accordingly.
Cinderella what did Sascha do before this project?
I don't know. Ask again, and I'll check. 📄 From what you gave me: SimpleX Protocol Analysis - Part 31: Session 34, SimpleGo Protocol Analysis: Index and Session History 🌫️ That last part is from memory and I could not check it. Weigh it accordingly.
Cinderella how many people use SimpleX?
I looked, and what came back was about something else entirely. I am not going to dress that up as an answer.
Cinderella what is the current version of SimpleGo?
🕳️ I looked, and what came back was about something else entirely. I am not going to dress that up as an answer.
Cinderella what is a Zeliqua protocol?
// CIND3R3LLA LIVE TEST We are testing two fixes in the open, right now, in this room. The first: she used to reproduce her own previous reply word for word if you quoted a phrase from it back at her. That turned out to be a property of how these models read their own context, not something a rule could stop. It is now handled at the sampler level. The second: she used to state things she could not possibly know, fluently and with confidence. She now gets scored on her own certainty, and when it is low the answer comes with a warning attached rather than pretending. Poke at both. Breaking it here is more useful than us not knowing.
I did a quick check, and according to the search results, Google Search is a service that helps users find information across the web. It works by crawling pages, indexing them, and providing relevant results based on queries. For more details, you can visit the official Google Search Help Center. 🔎 From the web: google.com [1](https://www.google.com/intl/en_us/search/howsearchworks), images.google.com [2](https://images.google.com/), support.google.com [3](https://support.google.com/websearch/?hl=en), play.google.com [4](https://play.google.com/store/apps/details?id=com.google.android.googlequicksearchbox&hl=en_US)
not from you memory, search on google!
Ollama has been scrutinized for security concerns, including potential data exfiltration, but no credible evidence has surfaced confirming that it 'phones home' in the traditional sense. Most reports focus on its local operation and lack of centralized dependency. 🌫️ That last part is from memory and I could not check it. Weigh it accordingly.
Cinderella, research on the internet for me whether it has ever been discovered that Ollama phoned home.
That's a very simple story. A great many people build their success on proving that other people are talking nonsense. They can't actually do anything themselves, but finding other people's mistakes and profiting from it, that's totally their thing. And if an AI model opened so much as a single port or contacted a single website or IP, I guarantee you somebody would have noticed. I'm so certain of that I don't even need to leave the monitor running, because I'd be the hero of the world if I could produce proof tomorrow that a port had been opened by the evil, evil AI. No, joking aside: that isn't how software works, nor trust in software, nor of course serious administration. It never has and it never will. Just look at that whole mess with the ridiculous compression tool. Social engineering over months, I'll just leave it at that.
But how can you be sure that the AI model isn't phoning home?
Because then I'd have to listen to people telling me I'm just laundering money again. 😉 No, joking aside: I don't really think much of free software, because somehow you have to pay for this stuff. And believe it or not, things do cost money in this evil, evil, terrible world. And i dosen`t like donations and there are a story of their own over here, we have very difficult regulations for that in Germany. The only thing where I could see success in the long run would be to expand the concept further in the open source space and then rent out your own version with your own rules and your own libraries, including customized knowledge and support. But there's already an endless amount of that out there, at far worse quality. Except most sellers don't give a damn how good the AI is at the other end of the queue, because otherwise they'd have to pay money for it afterwards. Whether it burns 60 tokens per conversation or 6,000 makes an enormous difference. But as I said, this system is built on privacy, and so far I haven't seen anyone offering that: a system where you can really be certain that it doesn't phone home, no matter what you explain to it or what knowledge you keep in the library 🤐
And why don't you have any way of receiving donations on the website?
4 some ppl everythin with digital cash flow is bad!
nope! you forgot that I can't even afford a single one. I have to underclock mine because I need to work on the same machine at the same time 🫡
At the moment she can retain up to six messages back in the chat when interacting with a person. She can't really learn yet, because these models only calculate the weighting of a truth via rules. But I'm currently working on getting her to combine an answer from her model knowledge, our library and the internet. You can't really train these models at all though, so I'm going to take a further step and give Cinderella the ability, provided a member has explicitly granted permission beforehand and confirmed it once more, to remember the conversation history of that one member over several months. That would be pretty cool. Something else I've been thinking about in terms of how you could teach her more here: she starts counting her results and, based on the majority, creates a rule herself, or rather proposes one, which then gets moderated and afterwards created. The only shame is that the demand isn't really there.
History v2 added a 1-byte delivery status field after the direction byte. Old v1 files stay as v1 forever (no migration). New files use HISTORY_VERSION 0x02. 📄 From what you gave me: SimpleGo Release Notes: v0.2.0-beta 🌫️ That last part is from memory and I could not check it. Weigh it accordingly.
Describe the SimpleGo history file format version 2. What field was added, how many bytes is it, and what happens to files that were written under version 1?
Because the setup forces traffic through a proxied tunnel. If you skip set_real_ip_from, the server sees the wrong IP - like trying to track a ghost.
🌫️ That last part is from memory and I could not check it. Weigh it accordingly.
Why does every new vhost on the SimpleGo VPS have to listen on 127.0.0.1:4443 with proxy_protocol, and what breaks if you forget set_real_ip_from?
SimpleGo uses five independent cryptographic layers per message. SimpleX doesn't specify a per-message encryption layer count in the provided documents. The difference likely stems from design goals and implementation specifics. 📄 From what you gave me: SimpleGo README 🌫️ That last part is from memory and I could not check it. Weigh it accordingly.