id
stringlengths 12
15
| title
stringclasses 100
values | content
stringlengths 0
8.16k
| prechunk_id
stringlengths 0
15
| postchunk_id
stringlengths 0
15
| references
stringclasses 100
values |
|---|---|---|---|---|---|
1811.00742#192
|
Rationality-proof consensus: extended abstract
|
Actors can't do much to influence the reward.
|
1811.00742#191
|
1811.00742#193
|
1811.00742
|
1811.00742#193
|
Rationality-proof consensus: extended abstract
|
They can
follow the protocol so it's back to their turn in a timely manner and they get the reward (that's
good).
|
1811.00742#192
|
1811.00742#194
|
1811.00742
|
1811.00742#194
|
Rationality-proof consensus: extended abstract
|
They can't jump the queue - we enforce a strict round-robin rule: everyone gets their turn.
|
1811.00742#193
|
1811.00742#195
|
1811.00742
|
1811.00742#195
|
Rationality-proof consensus: extended abstract
|
They can try to prevent other actors from joining - this is a topic on its own and we discuss it
elsewhere.
|
1811.00742#194
|
1811.00742#196
|
1811.00742
|
1811.00742#196
|
Rationality-proof consensus: extended abstract
|
Second, preventing blocks from being added is something a rational actor might
cheat for.
|
1811.00742#195
|
1811.00742#197
|
1811.00742
|
1811.00742#197
|
Rationality-proof consensus: extended abstract
|
Sending fake FREEZE messages doesn't work because it requires signatures from
other actors.
|
1811.00742#196
|
1811.00742#198
|
1811.00742
|
1811.00742#198
|
Rationality-proof consensus: extended abstract
|
The actor could try sending fake TIMEOUT messages to speed up eviction of a
leader that is being slow (or perhaps in a period of asynchrony).
|
1811.00742#197
|
1811.00742#199
|
1811.00742
|
1811.00742#199
|
Rationality-proof consensus: extended abstract
|
This also requires others to
cooperate but the argument is more subtle because some non-malicious actors may genuinely
observe a leader being slow.
|
1811.00742#198
|
1811.00742#200
|
1811.00742
|
1811.00742#200
|
Rationality-proof consensus: extended abstract
|
We use the clock rate assumption here: the next leader will reject
TIMEOUT messages if they arrive too early.
|
1811.00742#199
|
1811.00742#201
|
1811.00742
|
1811.00742#201
|
Rationality-proof consensus: extended abstract
|
Since the clock ratio is r, a message that arrives
before T/r is too early.
|
1811.00742#200
|
1811.00742#202
|
1811.00742
|
1811.00742#202
|
Rationality-proof consensus: extended abstract
|
If a rational actor sends at time T then it's guaranteed that the new leader
9
will accept it.
|
1811.00742#201
|
1811.00742#203
|
1811.00742
|
1811.00742#203
|
Rationality-proof consensus: extended abstract
|
In fact this is the earliest time the new leader is guaranteed to accept it, so it's the
optimal time to send (hence the algorithm is at equilibrium).
|
1811.00742#202
|
1811.00742#204
|
1811.00742
|
1811.00742#204
|
Rationality-proof consensus: extended abstract
|
Conversely, one might imagine
rational actors trying to prevent others from proposing a block.
|
1811.00742#203
|
1811.00742#205
|
1811.00742
|
1811.00742#205
|
Rationality-proof consensus: extended abstract
|
However, actors take turns and
they cannot end this round without suspect messages from other actors.
|
1811.00742#204
|
1811.00742#206
|
1811.00742
|
1811.00742#206
|
Rationality-proof consensus: extended abstract
|
Once we've shown the algorithm is a Byzantine Nash Equilibrium, for the rest of the argument
we can safely assume that rational actors will follow the protocol.
|
1811.00742#205
|
1811.00742#207
|
1811.00742
|
1811.00742#207
|
Rationality-proof consensus: extended abstract
|
The key point of the freezing
algorithm is that it will not result in two different blocks being decided because (a) if a block
could be decided by the previous round, then that block is discovered by the freezing algorithm
and proposed in the next round.
|
1811.00742#206
|
1811.00742#208
|
1811.00742
|
1811.00742#208
|
Rationality-proof consensus: extended abstract
|
(b) otherwise, the previous round will never decide a block.
|
1811.00742#207
|
1811.00742#209
|
1811.00742
|
1811.00742#209
|
Rationality-proof consensus: extended abstract
|
The
first point follows from the fact that the DECIDE message need 2f+1 signatures: at least one of
the signatories will be uncovered by the freezing algorithm and the block will be forwarded to the
next round.
|
1811.00742#208
|
1811.00742#210
|
1811.00742
|
1811.00742#210
|
Rationality-proof consensus: extended abstract
|
The second point follows from the symmetric argument that FREEZE is sent to 2f+1
actors.
|
1811.00742#209
|
1811.00742#211
|
1811.00742
|
1811.00742#211
|
Rationality-proof consensus: extended abstract
|
If all accept it before seeing a WRITE, then there are not enough left to sign the DECIDE
message so the previous round will not decide.
|
1811.00742#210
|
1811.00742#212
|
1811.00742
|
1811.00742#212
|
Rationality-proof consensus: extended abstract
|
This algorithm allows the system to pick a block and prevents malicious or rational actors from
preventing consensus or interfering with the turn order.
|
1811.00742#211
|
1811.00742#213
|
1811.00742
|
1811.00742#213
|
Rationality-proof consensus: extended abstract
|
Conclusion
In this paper we argue for algorithms that tolerate both rational (self-interested) actors, and
malicious (Byzantine) ones.
|
1811.00742#212
|
1811.00742#214
|
1811.00742
|
1811.00742#214
|
Rationality-proof consensus: extended abstract
|
These algorithms are called BAR-tolerant.
|
1811.00742#213
|
1811.00742#215
|
1811.00742
|
1811.00742#215
|
Rationality-proof consensus: extended abstract
|
When writing one, follow
these three steps: clearly define the utility function for the rational actors, prove the algorithm is
such that there is no benefit from unilaterally deviating (that is, it's a Byzantine Nash
Equilibrium), then prove the algorithm correct assuming the rational actors follow the protocol.
|
1811.00742#214
|
1811.00742#216
|
1811.00742
|
1811.00742#216
|
Rationality-proof consensus: extended abstract
|
We explain the gatekeeper attack, where members of a system selfishly decide to prevent
newcomers from joining.
|
1811.00742#215
|
1811.00742#217
|
1811.00742
|
1811.00742#217
|
Rationality-proof consensus: extended abstract
|
We sketch a BAR-tolerant blockchain protocol.
|
1811.00742#216
|
1811.00742#218
|
1811.00742
|
1811.00742#218
|
Rationality-proof consensus: extended abstract
|
It relies on a strict order
to decide who gets to propose a new block (so there's no need to race to solve a crypto puzzle)
and it relies on hardware ID tokens to make sure every computer is only represented at most
once as a block proposer.
|
1811.00742#217
|
1811.00742#219
|
1811.00742
|
1811.00742#219
|
Rationality-proof consensus: extended abstract
|
It also defends against the gatekeeper attack.
|
1811.00742#218
|
1811.00742#220
|
1811.00742
|
1811.00742#220
|
Rationality-proof consensus: extended abstract
|
The BAR-tolerant
approach is naturally also applicable to other blockchain algorithms.
|
1811.00742#219
|
1811.00742#221
|
1811.00742
|
1811.00742#221
|
Rationality-proof consensus: extended abstract
|
References
[1] Amitanand S.
|
1811.00742#220
|
1811.00742#222
|
1811.00742
|
1811.00742#222
|
Rationality-proof consensus: extended abstract
|
Aiyer, Lorenzo Alvisi, Allen Clement, Mike Dahlin, Jean-Philippe Martin, and
Carl Porth.
|
1811.00742#221
|
1811.00742#223
|
1811.00742
|
1811.00742#223
|
Rationality-proof consensus: extended abstract
|
2005.
|
1811.00742#222
|
1811.00742#224
|
1811.00742
|
1811.00742#224
|
Rationality-proof consensus: extended abstract
|
BAR fault tolerance for cooperative services.
|
1811.00742#223
|
1811.00742#225
|
1811.00742
|
1811.00742#225
|
Rationality-proof consensus: extended abstract
|
SIGOPS Oper.
|
1811.00742#224
|
1811.00742#226
|
1811.00742
|
1811.00742#226
|
Rationality-proof consensus: extended abstract
|
Syst.
|
1811.00742#225
|
1811.00742#227
|
1811.00742
|
1811.00742#227
|
Rationality-proof consensus: extended abstract
|
Rev.
|
1811.00742#226
|
1811.00742#228
|
1811.00742
|
1811.00742#228
|
Rationality-proof consensus: extended abstract
|
39, 5
(October 2005), 45-58.
|
1811.00742#227
|
1811.00742#229
|
1811.00742
|
1811.00742#229
|
Rationality-proof consensus: extended abstract
|
DOI:
https://doi.org/10.1145/1095809.1095816
[2] Bitcoin.org.
|
1811.00742#228
|
1811.00742#230
|
1811.00742
|
1811.00742#230
|
Rationality-proof consensus: extended abstract
|
2015.
|
1811.00742#229
|
1811.00742#231
|
1811.00742
|
1811.00742#231
|
Rationality-proof consensus: extended abstract
|
Some Miners Generating Invalid Blocks.
|
1811.00742#230
|
1811.00742#232
|
1811.00742
|
1811.00742#232
|
Rationality-proof consensus: extended abstract
|
https://bitcoin.org/en/alert/2015-07-04-spv-mining
10
[3] Bitcoin Wiki.
|
1811.00742#231
|
1811.00742#233
|
1811.00742
|
1811.00742#233
|
Rationality-proof consensus: extended abstract
|
Value overflow incident.
|
1811.00742#232
|
1811.00742#234
|
1811.00742
|
1811.00742#234
|
Rationality-proof consensus: extended abstract
|
https://en.bitcoin.it/wiki/Value_overflow_incident
[4] Vitalik Buterin.
|
1811.00742#233
|
1811.00742#235
|
1811.00742
|
1811.00742#235
|
Rationality-proof consensus: extended abstract
|
2014.
|
1811.00742#234
|
1811.00742#236
|
1811.00742
|
1811.00742#236
|
Rationality-proof consensus: extended abstract
|
On Stake.
|
1811.00742#235
|
1811.00742#237
|
1811.00742
|
1811.00742#237
|
Rationality-proof consensus: extended abstract
|
https://blog.ethereum.org/2014/07/05/stake/
[5] Vitalik Buterin.
|
1811.00742#236
|
1811.00742#238
|
1811.00742
|
1811.00742#238
|
Rationality-proof consensus: extended abstract
|
2014.
|
1811.00742#237
|
1811.00742#239
|
1811.00742
|
1811.00742#239
|
Rationality-proof consensus: extended abstract
|
Slasher: A Punitive Proof-of-Stake Algorithm. https://blog.ethereum.org/2014/01/15/slasher-a-punitive-proof-of-stake-algorithm/
[6] Miles Carlsten, Harry Kalodner, S.
|
1811.00742#238
|
1811.00742#240
|
1811.00742
|
1811.00742#240
|
Rationality-proof consensus: extended abstract
|
Matthew Weinberg, and Arvind Narayanan.
|
1811.00742#239
|
1811.00742#241
|
1811.00742
|
1811.00742#241
|
Rationality-proof consensus: extended abstract
|
2016.
|
1811.00742#240
|
1811.00742#242
|
1811.00742
|
1811.00742#242
|
Rationality-proof consensus: extended abstract
|
On the
Instability of Bitcoin Without the Block Reward. In
Proceedings of the 2016 ACM SIGSAC
Conference on Computer and Communications Security
(CCS '16).
|
1811.00742#241
|
1811.00742#243
|
1811.00742
|
1811.00742#243
|
Rationality-proof consensus: extended abstract
|
ACM, New York, NY, USA,
154-167.
|
1811.00742#242
|
1811.00742#244
|
1811.00742
|
1811.00742#244
|
Rationality-proof consensus: extended abstract
|
DOI:
https://doi.org/10.1145/2976749.2978408
[7] Dave Gutteridge, Japanese Cryptocurrency Monacoin Hit by Selfish Mining Attack, CCN
https://www.ccn.com/japanese-cryptocurrency-monacoin-hit-by-selfish-mining-attack/
(retrieved
2018-10-16)
[8] S.
|
1811.00742#243
|
1811.00742#245
|
1811.00742
|
1811.00742#245
|
Rationality-proof consensus: extended abstract
|
Micali, M.
|
1811.00742#244
|
1811.00742#246
|
1811.00742
|
1811.00742#246
|
Rationality-proof consensus: extended abstract
|
Rabin, S.
|
1811.00742#245
|
1811.00742#247
|
1811.00742
|
1811.00742#247
|
Rationality-proof consensus: extended abstract
|
Vadhan.
|
1811.00742#246
|
1811.00742#248
|
1811.00742
|
1811.00742#248
|
Rationality-proof consensus: extended abstract
|
1999.
|
1811.00742#247
|
1811.00742#249
|
1811.00742
|
1811.00742#249
|
Rationality-proof consensus: extended abstract
|
Verifiable random function.
|
1811.00742#248
|
1811.00742#250
|
1811.00742
|
1811.00742#250
|
Rationality-proof consensus: extended abstract
|
In
Proceedings of the 40th
Annual Symposium on Foundations of Computer Science.
|
1811.00742#249
|
1811.00742#251
|
1811.00742
|
1811.00742#251
|
Rationality-proof consensus: extended abstract
|
[9] Josiah Wilmoth, Bitcoin Gold Hit by Double Spend Attack, Exchanges Lose Millions, CCN
https://www.ccn.com/bitcoin-gold-hit-by-double-spend-attack-exchanges-lose-millions/
(retrieved
2018-10-16)
11
|
1811.00742#250
|
1811.00742
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.