We haven't moved up yet but when we do there's no stopping until $1
>Unironically the next link
>Check out the data marketplace for hardpp market.oceanprotocol.com
Easily my comfiest hodl of 2020
OCEAN
Colton Fisher
Levi Allen
Parker Flores
sold
Blake Richardson
you had all year to accumulate
let the fomo begin
Samuel Fisher
>let the fomo begin
Connor Watson
Ooooooooooooooooo
Angel Lewis
$1 end of month
Aiden Adams
Ocean Piss corp (discord):
yxmzgY9
Brody Thompson
Honestly I don't know. About a year ago I was building a company along the same lines as OCEAN (Without a blockchain layer). We have pivoted since then.
The issue they are going to run into is individual privacy. Where data pertains to European citizens, you are basically fucked. Privacy research / technology is really not at a place where you can support a full-blown marketplace for arbitrary datasets. Ocean is therefore reduced to basically training models.
Even when training models however, it's notoriously difficult to get right - and the definition of privacy they use (differential privacy) means that you basically need to know what features you want to train on the model a-priori. Furthermore, the addition of noise which happens in differential privacy reduces the accuracy of your model - which is bad from an objective model performance standpoint, but can also cause serious issues if there models are used to produce results in regulated industries (i.e. credit scores). Furthermore, the verdict on where differential privacy stands from a GDPR perspective is kind of blurred, but last I checked it counted as de-identified data and not anonymised data. You have to be very careful with de-identified data, you can't just sell it off (unlike anonymised data).
I hope they do well, the world needs something like this, but my money is on that it is a little early.
Sebastian Martin
they solve it by letting you run your algo on the data at the place where the data is stored
so for example you could run stats or train AI without destroying privacy or gdpr
Easton Roberts
Christopher Nguyen
I like your analysis. Do you also have any thoughts on PNK?
Michael Gomez
>We haven't moved up yet but when we do
Been hearing that for months. Also marketplace is shit every new set gets pumped and dumped regardless of what it is and you even have official partners pulling liquidity. Shame too, OCEAN could’ve been the one...
Jason Turner
Yeah and that sounds like a good idea right?
Well the truth is it's not that simple, even if you train the model where the data is stored (this is called federated learning), ML models by definition encode sensitive information in their weights (that's why you have the differential privacy layer which introduces the errors).
This means that companies will now need permission to process data for this reason.
Anthony Baker
>Been hearing that for months. Also marketplace is shit every new set gets pumped and dumped regardless of what it is and you even have official partners pulling liquidity. Shame too, OCEAN could’ve been the one...
you remind me of the ethereum complainers around ETH=$1
Ayden Kelly
>Well the truth is it's not that simple, even if you train the model where the data is stored (this is called federated learning), ML models by definition encode sensitive information in their weights (that's why you have the differential privacy layer which introduces the errors).
>This means that companies will now need permission to process data for this reason.
I mean maybe it's sensitive. If the dataset is sufficiently large I doubt anyone could make a claim about a ML weight leaking private data.
Jacob Wright
No idea. I know about data privacy. Not much else :P
Nolan Clark
They solve this by using secret network
Computations on encrypted data. Your never actually giving the data away. It remains encrypted the entire time
Both ocean and scrt are comfy holds
Joseph Davis
Absolute scam.
Even worse than HEX.
Henry Cook
>Computations on encrypted data.
I wasn't aware they did homomorphic application
Link?
Josiah Ortiz
seethe noceaner
Eli Rogers
Hi folks, does anyone use DMX token? Found on probit, looks legit. They have a modern staking system and workable tools for investors and traders. They have done private sale with 1mln: dymmax.com
Isaac Ortiz
Ayden Carter
Oooo
Thomas Reed
i sold 8000 link @ 15 and bought this
Evan Wood
>i sold 8000 link @ 15 and bought this
Noah Gray
Except the data you sell there actually comes from Streamr which is a 30 million marketcap lol, most underrated gem on this useless board. I'd rather 10x there than 2x with Ocean
Hudson Smith
Yeah, the only problem is fully homomorphic encryption is:
a) 6 orders of magnitude slower than plaintext computation
b) does not guarantee privacy at the output.
To give an example of b, imagine that you are performing a computation on encrypted text from data on your machine.
and the computation is, F = 0 + your_age. Even though the operation 0 + your_age is performed in ciphertext, the answer leaks private information about your age.
Dominic Smith
You seem to be a data analysis expert user
Tell me if you know of anything like this:
- An algorithm specification language
- Which lets you compute mathematical proofs over the algorithm
- Including a proof that the output destroys enough information that it can never be run backwards (thus preserving data point privacy)
Chase Peterson
>noceaner
I'm still not there. I'm trying to get rid of it.
This was one of the worst investments I've ever made.
Juan Price
noilers on sui watch
Day of the Barrel soon
then the only OCEAN relevant will be from you faggot's tears
Landon Moore
lmao how
where did you buy in