Hacker News Reader

Comment by StavrosK | Hacker News Reader

I've always wanted a trustless, decentralized, proof-of-work-based alternative to proprietary devices like Alexa and Google Home.

Oh, wait, I didn't. What I've wanted is some nice open hardware with a good ML model for recognizing speech. I don't know what a blockchain is doing here.

View on HN

And I do invite you to try out our voice platform, running 100% on-device on Raspberry Pi 3 or equivalent (and soon MCUs). We're on par with Amazon, Google et al. in terms of performance:

- https://makers.snips.ai/

- https://github.com/snipsco/snips-nlu

- https://medium.com/snips-ai/snips-nlu-is-an-open-source-priv...

This demonstrates nicely that the "we need the cloud to do good voice analytics" argument is baloney. Local voice recognition is perfectly do-able, especially on today's hardware with today's cheap storage. They need the cloud to do bulk surveillance.

Oh I know, I've tried it and that's why I like what you guys have been doing. The missing piece of the puzzle is a decent hardware kit, but putting a blockchain on any of this is completely unnecessary.

Totally agree that we need decent, open hardware, and that's why we're set on building it :) This will take some time, but in the mean time, we've launched a "Maker Kit" (https://makers.snips.ai/kit/) so that everyone can get started building voice assistants for their home that don't listen in.

The blockchain is a key ingredient in what we are trying to build, that is, 100% private AI assistants. We will detail this in our upcoming white paper.

Oh I'm sure that a blockchain is a key ingredient, I'm just not sure why your blockchain is a key ingredient. There are thousands of other coins you can use here instead of fragmenting the space even more and spending development effort on yet another blockchain.

I know that the goal of your blockchain is "cash in on that sweet sweet ICO hype", I just wish you either didn't or that you were more upfront about it.

How to determine if you need a block chain:

(1) Are you building a fully decentralized global ledger for use by multiple parties that do not share a common basis of trust?

If YES: Congratulations! You might need a block chain.

If NO: You do not need a block chain.

There are no more questions. :)

Exactly, and when there's a use case for a blockchain that isn't "building a trustless currency", I'd love to know, because everything else I've heard would be better suited by a database on a server.

Decentralized DNS and distributed CAs are possibilities, but those are quite close to a distributed trustless currency in that they apportion a finite resource or assign trust.

It started as we wanted to avoid using credit cards and thus having to identify users (the whole point of this new product is 100% privacy). The token will be user to run a TCR-like appstore on the consumer side, and to incentivize people to participate in federated learning developer side.

Why make your own token? Why not use an already-existing currency like Nano or something?

(It's a rhetorical question, I know the answer is "because we want to sell our tokens for a shitload of money")

Reply on news.ycombinator.com