Skepticism until product market fit.

nevent1qqsx5r6mnseh9rjrxunqpj3csm86c7gf7whywreqwc06xcmyr90lu8gppemhxue69uhkummn9ekx7mp0qy2hwumn8ghj7un9d3shjtnyv9kh2uewd9hj7zvz276

I tried to read it and didn't understand a thing.

It's built on top of the recently patched, bandwidth bottlenecked rendezvous routing system of Tor, that now has a CPU based proof of work for spam mitigation.

It essentially tries to make a full network out of the hidden service systems, which are already creaking from the attacks of the pre-PoW rendezvous protocol.

There was a messaging project called "torchat" some years back, that got abandoned because the Tor Project changed around their API too many times and broke the torchat author's code. It used tor hidden services to create overlay addresses, virtual addresses you might say, exactly in the way that Veilid is doing, more or less.

The problem is simple: bidirectional, telescoped onion routing is slow to build, and slow to recognise when it fails. Tor is not as slow or rigid as I2P, but it is in the same vein.

I'm working on a replacement for Tor, which dispenses with these bidirectional, telescoped channels, and uses the same technique used in Lightning, the Sphinx TLV onions, and uses a dual key, session based system for accounting bandwidth (and time) for sessions that are paid for using keysend/AMP style sender initiated lightning poyments, and a differently configured internal lightning network where there are no fees as the routers are being paid for relaying, not passing around payments.

The most important difference is that Indra hidden services don't have a bottleneck in the same way as Tor hidden services. Tor hidden services are backed up by 6 two directional 3 hop paths that meet and form a rendezvous, meaning that at any given time, bandwidth capacity and thus latency are highly constricted, a hidden service might have a 2.5Gbit pipe but most of the relays are 100 or 300, and everything necks down to that, and the congestion that can bring, etc etc. Indra hidden services basically are designed such that most of the time, the full bandwidth of a server can be used.

And overall, the biggest difference is this design has nothing to say about congestion control, or incentivising running of the software that enables it. Indra uses milliSats as a tool for making spam costs linearly scale, ie, making it uneconomical. Tor has no such thing, and that suits No Such Agency very well, who are probably running half of the Tor network by now.

🚩 "Veilid has no profit motive, which puts us in a unique position to promote ideals without the compromise of capitalism." Big reason Bitcoin is kicking ass is because of profit motive.

Skeptical.

In theory, with anonymity software, diversity is an attack (due to fragmentation of the anonymity set). In simple words, if it's not Tor, it's an attack on Tor.

Not 100% sure that this is what the world needs right now.

Cautiously bullish