It's been a while since I've posted on Hive. A lot has happened in the past year so I'll recap. Remember that project Tau I kept posting about? It's finally at test net. In fact, if you're reading this, it's days away from 1.0.
Andrei K, the developer of testnet, just did some updates. You can find the testnet here: https://github.com/IDNI/tau-testnet
I joined Steemit roughly 10 years ago, and during that time TauChain as it was called, was just a whitepaper. This was before Hive split off into it's own faction. Many of the topics you're reading about today in the news (agents, formal verification, alignment), are what myself and others talked about a decade ago. The problem is, good things can sometimes take a long time to build.
Now we are at a time where we need exactly what Tau is about to bring. Governable artificial intelligence. Sovereign aligned agents. This is contrary to what you see in the mainstream right now, where the agents are corporate controlled, and AI is being positioned as something to compete with or even replace the human. That isn't what I had in mind when I envisioned agents 10 years ago.
On the other hand some events are playing out exactly as I expected. The many smart contract hacks, the technological unemployment, the sharp divisions around how the technology should be used. Some ideas from 10 years ago have not yet been built out, and are now close to being feasible, such as Network States, or cyborgization. 10 years ago these ideas sounded crazy, now they sound inevitable.
The next few years will be both challenging, and also have opportunities to build great digital communities. That is something I intend to participate in, and that is why now is the time for me to become active again on Hive. I'll be sharing my own projects, some of which will be built on Tau, some which will be built for Hive.
Tau Net, the former Agoras?
I remember, that project was full of promises and endless delays and then magically used the hype word AI to transform itself into an AI company. Like Allbirds. I don´t buy it.
PS: You have a quite unhealthy KE ratio. Upvoting you is like throwing Hive directly into the sink. Just saying, in case you want to get real here.
It was always an AI company. Agoras was talking about AI before OpenAI released ChatGPT. It's just now AI is fashionable, but Tau is not the same kind of AI as LLMs. Tau is the other side of AI, the kind of AI that can reason, logic, etc. What is K/E ratio? That didn't exist when I posted in the past.
I looked it up, it seems to be a recent 2024 metric of some sort. I'll learn how that works.
But what did Tau Net achieve in all these years, apart from great concepts?
In a fraction of the time companies like OpenAI and Anthropic generated bio. of revenues (and even more losses but that´s another story).
Why don't you go and see for yourself? The testnet is running, the website explains: https://tau.net
Part of doing things right, takes longer. Other blockchains have been hacked for hundreds of millions of dollars, sometimes more than once. I agree, it's taken a long time, but if you didn't have to hold anything during that time, you didn't have to pay the opportunity cost, so you'll get to use it soon. Myself, I'll admit I had to pay the 10 year opportunity cost, but we can't all win.
The website is nice but is there a single functioning use case where TAU made a difference? I have seen too many of those convincing project sites of tokens that are all gone by now.
It's testnet so I can't lie and claim it's making a financially measurable difference. But I can say the technology makes a difference. If you heard of "Constitutional AI", from researchers at Anthropic, or AI companies saying that term, it was actually us who invented it. Ohad in particular invented Tau, while myself and others invented or at least popularized the idea of a "smart Constitution" or AI Constitution. The difference between our idea of Constitutional AI and Anthropic's idea, is ours is a decided by the community. The opinion map, the million person discussion, all of this was about creating the rules to govern AI, and only Tau has the ability to do this.
Ohad Asor did the research which you can find in Theories and Application of Boolean Algebra, on the website. This research is what allows for Tau Language which has unique patented properties allowing for a formal constitution to be created. Unlike Anthropic's constitution which has to be enforced centralized, by a company, where you and I have no input on what it says, and Claude isn't guaranteed to follow it, with Tau, I can say that agents are mathematically guaranteed to follow the rules. This means Tau Network is the only network that exists where agents are governable. Tau Language is the only language that exists that allows for true Constitutional AI.
That's the breakthrough. Now we can look at some other stuff which isn't quite unique to Tau but is still valuable. Tau Language is correct by construction, any decentralized app is a formal specification, which means the sort of bugs where hundreds of millions or billions of dollars are lost, are mathematically impossible. Solana, Ethereum, these suffer vulnerabilities every few weeks, and now they are terrified of Mythos because of how many vulnerable smart contracts their network has.
Tau Net apps written in Tau Language will be invulnerable to the sort of hacks that commonly impact those other chains or even Hive. The only other chain I know of similar to this is Tezos, which also has never been hacked as far as I know. This is important because if we are going to build a decentralized economy, we don't win if people keep losing money from hacks, the whole DeFi loses legitimacy when this happens, and it's hurting everyone.
I hope that summarizes some of what Tau Net can do that no other technology can do. And I hope that shows you the value in the agentic era, where it's all about governing agents not just setting them to create slop and chaos.
Thanks for your insights.
It seems you are under the assumption that a decentralized governance is a value for the Tech elite (those who are behind the current LLMs). I am a hard core Hivians for over 9y, so this assumption is in my veins, too. But the sad reality is that centralization brings $$$, not decentralization (just look at the Hive price, but that has other reasons too). Maybe OpenAI and Co. do not want a decentralized and transparent, constitutional governance. Who in fact wants to follow decisions by the community (where 90% are dumb and easily manipulated)? Even in our socalled democracies, the elections are mostly fake and make people believe to have a choice (if they would really change things they are quickly revoked, like 2024 happened in Romania).
Sorry, I didn´t want to derail into politics, but great technological concepts and economic success are 2 completely different things.
OpenAI started in 2015 and released something around 2023. That's 8 years. Anthropic didn't exist, but sure they beat us to the punch. If you want to say other AI came faster, it's true.
But it's also not the same kind of AI. So it's not a threat. And OpenAI is not open or decentralized. You're relying on a centrally controlled AI, so when it comes to blockchain, Tau is one of the rare cases where we have something real. The only other alternatives are Bittensor and SingularityNET, and those work entirely different, less decentralized, different use cases.
Regarding K/E, I think we can come up with better metrics. But if I remember curation is mostly done by bots. Bots can't review content for quality, however when we attach AI to bots, perhaps they can review for quality. I think this platform can be renewed.
I can assure you I am not a bot, I curate manually for a good part and hence I use also some metrics, e.g. the KE to decide.