You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+86-77Lines changed: 86 additions & 77 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -6,6 +6,8 @@
6
6
7
7
# BetterSign
8
8
9
+
[(https://www.youtube.com/watch?v=LxU4wG4ryFo)]
10
+
9
11
## Introduction
10
12
11
13
BetterSign (`bs`) is a new signing tool designed to use provenance based
@@ -32,29 +34,34 @@ therefore lack the ability to prove the history of control and modification of
32
34
any digital data, let alone cryptographic keys. With the creation of
33
35
blockchains and distributed consensus popularizing the idea of immutable
34
36
records of transactions over time, we've learned the value of maintaining logs
35
-
to document the provenance of data over large spans of time despite the
36
-
unbounded memory requirement that results. In a world where there is old data
37
-
signed with old keys, there must be some cryptographically verifiable record
38
-
preserving and linking old keys to the new keys; provenance logs are design to
39
-
be the simplest and most decentralized solution for that.
37
+
to document the provenance of data despite the associated unbounded memory
38
+
requirement. In a world where there is old data signed with old keys, there
39
+
must be some cryptographically verifiable record preserving and linking old
40
+
keys to the new keys; provenance logs are design to be the simplest and most
41
+
decentralized solution for that.
40
42
41
43
To finally improve the global PKI system, it seems logical to start from
42
44
scratch and construct an identity solution based entirely off of a provenance
43
45
logging structure. It also is apparent that identity transactions are either
44
-
1-party or 3-party transactions and therefore do not require the distributed
45
-
consensus necessary for trustful 2-party transactions. This opens the door for
46
-
provenance logs that grow in trust in several ways. They can accumulate 1st
47
-
party self-attestations along with proofs of work (i.e. content creation of all
48
-
kinds or verifiable acts of service). They may also record references to 3rd
49
-
party corroborating attestation sources for realtime, late-binding verification
50
-
from multiple trustworthy societal institutions or organizations. In the end,
51
-
this solution is very good at overcoming the analog-to-digital problem of
52
-
encoding verifiability. The security rests in the statistical improbability of
53
-
corrupting and/or falsifying proof from an increasing number of trustworthy
54
-
insitutions while also making verification time-sensitive and responsive to
55
-
shifting facts on the ground. This corroboration based security model gives
56
-
statistical assurances of what is true and is the native model for provenance
57
-
logs.
46
+
1-party ("trust me, bro") or 3-party ("they vouch for me") transactions and
47
+
therefore do not require the distributed consensus necessary for trustful
48
+
2-party transactions. This opens the door for provenance logs that grow in
49
+
trust in several ways:
50
+
51
+
1. They can accumulate 1st party self-attestations along with proofs of work
52
+
(i.e. content creation of all kinds or verifiable acts of service).
53
+
54
+
2. They may also record references to 3rd party corroborating attestation
55
+
sources for realtime, late-binding verification from multiple trustworthy
56
+
societal institutions or organizations.
57
+
58
+
In the end, this solution is very good at overcoming the analog-to-digital
59
+
problem of encoding verifiability. The security rests in the statistical
60
+
improbability of corrupting and/or falsifying proof from an increasing number
61
+
of trustworthy insitutions while also making verification time-sensitive and
62
+
responsive to shifting facts on the ground. This corroboration based security
63
+
model gives statistical assurances of what is true and is the native model for
64
+
provenance logs.
58
65
59
66
Provenance logs are a form of time-based log with the added feature of
60
67
cryptographically enforced write priviledges which may be delegated and
@@ -72,32 +79,34 @@ README][PROVREADME].
72
79
73
80
When discussing distributed systems we speak of networks of peers connected
74
81
together with links. A network consists of peers with links that reference
75
-
other peers. The links are an identifier that may reference the peer, a service
76
-
provided by a peer, or data stored by a peer. A link does not necessarily imply
77
-
an active network connection but does imply that one will be created when the
78
-
link is used to execute the distributed functions of the network.
82
+
other peers. The links are an identifier that either reference a peer, a
83
+
service provided by a peer, or data stored by a peer. A link does not
84
+
necessarily imply an active network connection but does imply that one will be
85
+
created when the link is used to execute the distributed functions of the
86
+
network.
79
87
80
88
All distributed systems are chaotic in nature meaning that the range and trends
81
89
in network behavior observed over time are impossible to predict from the
82
90
current conditions. However, distributed systems may be categorized into two
83
91
buckets based on their long-term stability and resilience in the face of the
84
92
corrosive effects of time. One category—*unstable* systems—are those that exist
85
-
at a point in timebut due to the design characteristics dictating peer and
86
-
link behavior they are *not* biased towards stability and never trend towards
87
-
*metastability*. These unstable systems often have many small localized
88
-
networks of peers but they never seem to conglomerate into a single long-term
89
-
network. You never get *THE* network—as in *THE* World Wide Web—arrising
90
-
spontaneously from *unstable* preconditions. The primary example of an
91
-
*unstable* network is the global identity "Web of Trust". Despite decades old
93
+
at a given point in time—say t0—but due to the design characteristics of the
94
+
peer and links they are *not* biased towards stability and never trend towards
95
+
*metastability*. Unstable systems often have many small localized networks of
96
+
peers but they never seem to conglomerate into a single long-term network. You
97
+
never get *THE* network—as in *THE* World Wide Web—arrising spontaneously from
98
+
*unstable* preconditions. The primary example of an*unstable* network is the
99
+
various attempts at building a global "Web of Trust". Despite decades old
92
100
standards and long-established market conditions, we still do not have *THE*
93
-
Web of Trust. This is likely due to the characteristics of pubkey links.
101
+
Web of Trust. This is likely due to using pubkeys as the identifiers and links
102
+
in the network.
94
103
95
-
The other category—*metastable* systems—are those with peer and node
96
-
characteristics that bias the chaos towards the accretion of a single, stable
97
-
network. These *metastable* networks start with a set of preconditions that
98
-
make *THE*network innevitable from the common usage patterns. The primary
99
-
example in this category is *THE* World Wide Web. This is also likely due to
100
-
the characteristics of URL links in the system.
104
+
The other category—*metastable* systems—are those with peer and link
105
+
characteristics that bias it towards the accretion of a single, stable network.
106
+
These *metastable* networks start with a set of preconditions that make *THE*
107
+
network innevitable from the common usage patterns. The primary example in this
108
+
category is *THE* World Wide Web. This is also likely due to the
109
+
characteristics of URL links in the system.
101
110
102
111
One key insight that comes from comparing pubkey links with URL links is that
103
112
pubkeys links can only be in one of two states—*valid* or *invalid*—while URLs
0 commit comments