You can also run old, slow, under provisioned, servers on the existing clear net too.
The question is, how fast is ‘fast enough’. Unresponsive nodes can be shunned. Defining what is unresponsive depends on what the goals are.
It is perfectly possible that there will be both very slow and very fast forks of Autonomi, should there be a demand. I suspect the sweet spot, will be somewhere in the middle, as dictated by user demand.
Also bear in mind that we have loads of caching on the clear net. Edge caching, server caching, browser caching, etc. Some of those are in place with Autonomi, but many aren’t. There is lots of headroom to meet higher performance goals.
Indeed, I’m still refining AntTP and getting more performance out of it. Even the current published version is pretty quick for IMIM, even through a shared gateway.
Spoiler: The next AntTP includes an async command queue, which makes IMIM article publishing feel instant and reduces the impact of pointer to chunk resolution.
We’re streaming movies directly from this thing now, dropping huge parts of the network when thousands of nodes shut off at once, with no detectable data loss. It really is amazing what we already have. I don’t think any other distributed data network has been as resiliant as this one. It still has its quirks, which is to be expected. It will only get better from here.
All that said, from what is being done in saorsa, it seems to me that autonomi really is just a piece of the final puzzle:
Local data first
Saorsa P2P for high performance data transfer and real time comms
Saorsa DHT for shared and private ephemeral/mutable data
Autonomi for shared and private persistent data
The 4 put together builds a full system that can actually compete with the legacy big tech stacks. Autonomi by itself is a great breakthrough for the persistent storage piece, but it isn’t the whole stack.
This is my opinion of course, but based on the work being done in the saorsa libs and autonomi, with the intent being to combine these together at some point, this is the most logical way to leverage these libraries’ strengths to build a complete platform.
I haven’t yet seen anything in the saorsa stack relating to the shared compute side of the equation. My guess is that will come after the data (and money) side is solved. Creating an autonomous P2P market place for encrypted general purpose compute (and storage) run on consumer grade hardware will be huge. And ‘consumer grade hardware’ in 10 years will be super compute clusters in personal robots working in kitchens and autonomous vehicles parked in garages. The trillions of $$$ being spent on these data centers today will pale in comparison to the distributed compute capabilities of billions of robots and autonomous vehicles.
Saorsa and Autonomi are parts of this stack that will enable us as humans and our personal devices to compete against the big tech monopolies shoving more and more GPUs into warehouses by having this distributed platform ready as the hardware rolls out into peoples’ homes. I just wish we had more people to help develop it…
This really sounds amazing. I haven’t personally experienced it yet but am taking your word for it. I have been waiting for these steps in progress for a very long time with many wishes.
The main wish is for it to integrate a digital currency that has all the best pieces. If it was the perfected money system and was ‘remote village user friendly’, just imagine how that could drive adoption! Imagine someone installing something for secure, fast, decentralized, finance and getting low-cost, unlimited, secure, fast, decentralized storage, comm, etc., as an added benefit. There are just so many synergistic gains for individuals and human advancement possible.
I’ve opened a topic on the new money system recently to begin introducing what is needed and how I (my past groups) propose it could work. You all should check it out.
Eating inulin reverses fatty liver disease by promoting gut bacteria that eat fructose (the primary cause of fatty liver disease). Inulin is a fructose polymer carbohydrate (a chain of fructose molecules) that our bodies can’t digest or break down, but some bacteria can. By enhancing the number of such bacteria in our gut, we naturally reduce the fructose in our gut from sugar as well. There are many vegetables that contain inulin, I won’t list them here, but you can easily find and learn about them via a search engine.
This also means training can proceed much faster and with less capable hardware. This is incredible and perhaps marks a major leap forward for AI development.
Will sure be interesting to see if it is good for coding or only playing games. If it is as good at coding Python as playing Pong then that will be interesting.
I don’t know what all the tests meant, I am in lack of knowledge of what some of them mean.
Would be great with a good ai that could work on a single Radeon 9070 xt.
I’ve looked into a bit more and it seems they are training the model to do recursive analysis of it’s own output - and not using recursion during training.
This implies that this method won’t work well with large models (as recursive inference with a large model will take significantly more processing and memory) and likely why they started testing this method with very small models.
Ideally, using some mixture of experts system with several different very small models may yield an interesting result.
Such a stack could certainly become a complete system, but I would like it to serve people first and foremost.
A long time ago, @dirvineassumed using an autonomous network for communication between self-driving cars. The problem is that they will belong to technology corporations (this will be dispassionate so-called on-demand transport), rather than to individuals (in practice, they will not be private property). Another problem is that autonomous cars are computers on wheels, in which transporting people will be only one of their secondary functions, which means that they will mainly be used for edge computing and, if necessary, will use the aggregated computing power of the autonomous network, and will also act as stabilisers for the energy network (hence politicians insist on a plug-in battery as the definition of an EV).And what worries me most is that this will not be a human-centric solution, but rather the use of these super clusters for distributed computing capabilities that will power a system serving big tech companies.
Fortunately, this will not happen in 10 years, as many countries will not agree to test autonomous cars on public roads for obvious reasons.
“In 2016, journalist Del Bigtree issued a challenge to the head of infectious disease at one of the most prestigious medical institutions in the world: conduct the most thorough vaxxed vs. unvaxxed study that has ever been done. The expert took up the challenge and ran the study to prove Del wrong. That study never saw the light of day… until now.”
Is this April 1st? Scientists invent Ice-9? … No, they’ve gone well past the fictional Ice-9 … we’re at Ice-21 – Ice XX1, but with a similar property to the fictional Ice-9. Stability at room temperature. Fortunately, unlike the fictional ice-9 of Kurt Vonnegut’s book “Cat’s Cradle”, this Ice-21 doesn’t create a chain reaction and ‘freeze’ all the water it touches. But just wait, Ice 22 may be on the horizon! lol Probably none here have read the book and have no clue what I’m talking about. It’s the only book that ever scared the sh!t out of me - very worth a read.
