Zoom Logo

Secure Data Storage - WG - Shared screen with gallery view
Dmitri Zagidulin
14:41
https://hackmd.io/V7UENLVFRKOWXJzy5316wA
Manu Sporny
16:00
haha
Manu Sporny
16:04
that was awesome
tobiaslooker
17:51
Please add yours if it is not listed
Sze (Z) Wong
17:58
Hi Everyone. This is the link to my company: https://www.zerionsoftware.com
Juan Caballero
18:10
encrypted queens mile ;)
tobiaslooker
18:18
;)
Juan Caballero
21:25
anyone have the hack md link handy?
Dave Longley
21:39
https://hackmd.io/V7UENLVFRKOWXJzy5316wA
Juan Caballero
21:44
thx
Dave Longley
24:10
we have a couple of people with their hands up
Manu Sporny
24:40
ooh, that one should go down on the list Juan
tobiaslooker
24:46
Yeah +1!
Daniel Buchner
24:54
confidential only pertains to the data that you don't want publicly available
Daniel Buchner
25:21
At the decentralized app level, you have fully confidential/private, shared with a sub-group, and fully public
tobiaslooker
25:24
Does confidential just mean a constrained audience?
Daniel Buchner
25:39
Could, but what about data that is fully public?
Manu Sporny
26:11
Daniel -- the difference here might be choice
Daniel Buchner
26:20
We need to stop thinking in terms of everything private - it's just a gradient for every object
Juan Caballero
26:20
mind BLOWN
Daniel Buchner
26:27
sure
Daniel Buchner
26:32
every object is a choice
tobiaslooker
26:43
Yeah I think choice is the distinction
Evan Tedesco
26:46
Zero trust data vault
Evan Tedesco
26:56
Lol took the words from my mouth
Manu Sporny
27:02
good point, Sze!
Juan Caballero
27:07
hmmmmm
Jim St.Clair
27:23
Zero Trust Data Fabric :)
Dave Longley
27:26
q+
Juan Caballero
27:35
haha zerotrust cloud storage :)
Daniel Buchner
28:11
Or deleting it all lol
Juan Caballero
28:32
or moving your server to a sanctioned country hehe
Jim St.Clair
28:40
+1 Juan
Manu Sporny
28:42
q+ to point out challenge/difficulty/elephant
Dave Longley
28:47
reduced trust isn't as catchy :)
Dave Longley
28:58
neither is "limited trust"
Daniel Buchner
29:08
Should use the most differentiated concept to lead it
Michael Shea
29:12
sort of trusty?
Daniel Buchner
29:24
because that will be the thing most devs and users see and interact with, plus it's less hazy
Adrian Gropper
29:26
Confidential or Encrypted are just too specific to be useful
Juan Caballero
29:41
they share did anchoring
Dave Longley
29:44
i think more specific is ok for the lower level primitive (the Encrypted Data Vault)
Dave Longley
29:53
the higher level thing should be less specific
Juan Caballero
29:57
but its hard to sell that to people who don't care what a did is
Dmitri Zagidulin
30:33
q+
Manu Sporny
30:40
Trusty Hubs
Dave Longley
30:42
much of "hubs" is not about servers, so we shouldn't prefer that in the top level name, IMO
Dave Longley
31:25
IMO, resource servers are already a dime a dozen in various specs, it doesn't differentiate enough
Michael Shea
31:40
Instead of trying to find a short set of words that describe what it is, would it make sense to just pick a ‘word’, a la ‘Aries’
tobiaslooker
32:12
For instance you could argue Dropbox is a secure data storage solution
Michael Shea
32:13
is there a mythical or legendary thing that fits?
Dave Longley
32:36
i think going with something like "Aries" makes it sound too much like a product and not a technology
Daniel Buchner
32:39
+1
Jim St.Clair
32:40
+1
Manu Sporny
32:41
+1 to exclude the word "Secure" -- it's misleading
Sze (Z) Wong
32:41
+1
Dmitri Zagidulin
32:47
PROPOSAL: Exclude the adjective “Secure” in name candidates
Dmitri Zagidulin
32:50
+1
Manu Sporny
32:51
+1
Michael Shea
32:51
+1
Dave Longley
32:52
+1
tobiaslooker
32:52
+1
Juan Caballero
32:55
+ to exclude
Evan Tedesco
33:08
+1
Adrian Gropper
33:11
0
Manu Sporny
33:13
yesss, yessssss, let the haaate flow through you (for the word "Secure")
Daniel Buchner
33:31
Encrypted also problematic
Juan Caballero
33:32
Insecure but Confidential Trusty Hubs
Jim St.Clair
33:43
+1 "trusy"
Jim St.Clair
33:46
trusty
Daniel Buchner
33:52
Gives the impression that everything is masked from everyone
Manu Sporny
33:54
Hard no to "Decentralized"?
Dave Longley
33:55
encrypted is literally true for the low level primitive, i don't see it as problematic in the same way "secure" is
Daniel Buchner
34:07
Mesh better than decentralized
Juan Caballero
34:10
same, not a huge fan
Jim St.Clair
34:22
Data Mesh is a thing too
Juan Caballero
34:34
I prefer "confidential" or "opaque" to define that it's hard to learn or deduce from watching it (i.e. traffic analysis, correlation, etc)
Daniel Buchner
34:42
Imagine calling Windows Pro Windows Full Disk Encryption
Dmitri Zagidulin
34:52
q+
Juan Caballero
35:00
Clippy Store
Dave Longley
36:02
obviously someone can decrypt it at some point, or it's not useful :)
Evan Tedesco
36:08
kill “Pubsubhubbub hubs” if anything it should be Pubsubhubbubs.
Jim St.Clair
36:10
That'll be a "no" to your "no"
Daniel Buchner
36:35
Decentralized doesn't === multiple copies
Daniel Buchner
36:37
imo
Juan Caballero
36:39
Evan, I put that one, snarkily, before the "no snarky names" rule got written
Juan Caballero
36:47
i'll see myself out
Evan Tedesco
37:01
I’ll allow it.. :)
Daniel Buchner
37:38
The name can be something other than a single attribute of the product/output
Manu Sporny
38:22
RCV them separately?
Dave Longley
38:28
self-sovereign storage SSS/S^3 (high level)
Manu Sporny
39:12
agree that marketing is important
Dave Longley
40:02
Pumpernickel
Jim St.Clair
40:37
oh yeah, a code name
Michael Shea
41:12
Rumplestiltskin?
Manu Sporny
41:29
it does imply larger scale
Jim St.Clair
41:40
"Mesh" is kinda copyrighted
Juan Caballero
41:47
^ IMPORTANT
Jim St.Clair
41:51
Fabric and Lake have architectural implications
Juan Caballero
41:53
is it associated with one company?
Jim St.Clair
41:57
yes
Jim St.Clair
42:02
https://www.youtube.com/watch?v=52MCFe4v0UU
Juan Caballero
42:10
sounds like a big negative :/
Jim St.Clair
42:24
https://www.thoughtworks.com/radar/techniques/data-mesh#:~:text=Data%20mesh%20is%20an%20architectural%20paradigm%20that%20unlocks%20analytical%20data,intensive%20applications%20across%20the%20organization.
Jim St.Clair
42:46
"Fabric" TMK is not
Juan Caballero
42:52
tbqh I tend to avoid "self-sovereignty" because it's too close to a trademark IMHO
Juan Caballero
43:10
for my very anti-enclosure preferences :D
Juan Caballero
43:30
thanks kaliya
Manu Sporny
43:44
Everybody, make all the decisions, now!
Juan Caballero
44:00
+1
Sze (Z) Wong
44:02
+1
Manu Sporny
44:05
+1
Jim St.Clair
44:07
too much pressure!!
Daniel Buchner
44:08
+1
Michael Shea
44:10
+
Dave Longley
44:11
+1 sure why not
Jim St.Clair
44:12
+1
tobiaslooker
44:13
+1
Adrian Gropper
44:19
+1
Dmitri Zagidulin
44:23
PROPOSAL: Ranked choice voting will be separately for/against Nouns and Adjectives separately
Michael Shea
44:30
roll of dice
Dave Longley
44:44
i don't think we've considered Storage McStorageFace yet
Michael Shea
44:51
+1
Dmitri Zagidulin
45:26
@Dave - added to the list! can’t pass that up.
Michael Shea
45:54
just remember the last time Boaty McBoat Face was on a list…..
Daniel Buchner
46:46
ZKP and ZTD
Dave Longley
47:04
ZTD sounds like what you get from doing too much ZKP
Manu Sporny
47:18
Marketing -> Cryptography
Manu Sporny
47:25
as we do in this group...
Dmitri Zagidulin
47:26
https://docs.google.com/presentation/d/1QEHSs4XJ05yQl2mvpiqbM80-MySxlVI9cNDLPq_XkkY/edit#slide=id.g8a2289ca98_0_0
Dave Longley
51:50
adrian's hand is up
Kaliya Identity Woman
52:02
I know
Dave Longley
52:05
ok :)
Manu Sporny
55:05
hmm?
Manu Sporny
55:11
oh, right
Manu Sporny
55:24
The way Dmitri said it made it seem strange...
Manu Sporny
55:37
but I think I get what he was saying.
Dave Longley
57:28
q+
tobiaslooker
57:43
Could you give a concrete example?
Dave Longley
57:45
q-
Jim St.Clair
58:16
+1 Tobias, sorry this feels like we're poking at what technology we think we want
Daniel Buchner
01:00:12
Why do we have to have two :/
tobiaslooker
01:00:14
q+
Daniel Buchner
01:00:34
having two separate permissioning systems sounds like a tall task for a single spec
Jim St.Clair
01:00:36
authorization/attenuation OR dynamic consent
Manu Sporny
01:00:53
Daniel - we don't have to have two -- we want to have one.
Jim St.Clair
01:00:55
schema-based consent
Manu Sporny
01:01:16
but the requirements may make it so we need more than one mechanism.
Manu Sporny
01:01:20
(and I hope we don't go there)
Dave Longley
01:02:05
i don't mean that the permission mechanisms need to be different, but i think there are entirely different types of policy that would need to be supported -- we don't need all the smarts at the server that stores the encrypted data
Dave Longley
01:02:48
adrian is looking for very rich policy processing -- which doesn't belong, IMO, on dumb storage servers that are just serving up encrypted data (that they cannot decrypt)
Manu Sporny
01:03:23
lols at the "read_page" action also allowing the "and delete comments" for Facebook.
Manu Sporny
01:04:00
Oh, OAuth 2.0 -- you're such a rascal!
Juan Caballero
01:04:16
delete your own comments on the page you're reading?
Manu Sporny
01:04:29
/me deletes Juans comment by reading it.
Manu Sporny
01:04:33
oops
Juan Caballero
01:04:41
ouch
Juan Caballero
01:04:44
tough crowd!
Jim St.Clair
01:04:50
harsh
Manu Sporny
01:04:52
I didn't meant to!
Manu Sporny
01:05:15
I was just using OAuth 2.0 -- I was young and a confused deputy.
tobiaslooker
01:05:21
Haha
Juan Caballero
01:06:01
it's ok, I'm a byzantine general anyways. i'm used to a little confusion.
Michael Shea
01:06:17
and now its parity
Manu Sporny
01:06:19
haha
tobiaslooker
01:06:48
I suggest we try and define things like the following resource, action and permission
Jim St.Clair
01:06:56
@Juan, just don't be orthogonal
Manu Sporny
01:07:30
+1 to that -- feels useful... meta authz requirements
Adrian Gropper
01:07:33
Use this for GNAP https://datatracker.ietf.org/doc/draft-richer-transactional-authz/?include_text=1