<redlizard>
AniSkywalker: The "alive" constraint makes this apparently impossible. There's no cryptographic distinction possible, as far as I can see, between a token based on votes that were alive at the time of granting the token, and a token based on votes that were not alive when granting the token.
<AniSkywalker>
redlizard I see. The main thing I want to do is make sure someone can't vote for someone else ten times.
<AniSkywalker>
The multiple people part is less of a problem.
<AniSkywalker>
However, it does need to reset upon issuance of a point.
<redlizard>
That too is problematic.
fleeky_ has quit [Ping timeout: 268 seconds]
<redlizard>
Cryptographic statements of a decision are not easily retracted.
MikeFair has quit [Ping timeout: 260 seconds]
galois_d_ has quit [Remote host closed the connection]
galois_dmz has joined #ipfs
<redlizard>
AniSkywalker: What is possible, as an approximation, is a point granted based on (a) N recent votes, plus (b) consensus between M peers that these N votes are recent.
<AniSkywalker>
That could work.
stoopkid has joined #ipfs
<redlizard>
This requires those M peers to be online at more or less the same time. And it might be vulnerable to malicious peers among the B groun.
<redlizard>
*group
<AniSkywalker>
redlizard what if I had a blockchain? Would that remove the online part, or do all peers need to be online to confirm the recency?
cemerick has joined #ipfs
leeola has quit [Quit: Connection closed for inactivity]
<redlizard>
AniSkywalker: I haven't really thought this part through, so here be dragons... but I imagine you'd require those M peers to stamp the N peers, and once they have all done that, stamp the set of N+M.
sirn has quit [Ping timeout: 260 seconds]
<redlizard>
So the M don't need to be online at the same time. But they do need to stay online frequently until the point is granted.
<AniSkywalker>
That's fine, since the goal here is to make it harder to commit a point.
<AniSkywalker>
To be clear, a point in this instance is bad.
sirn has joined #ipfs
<AniSkywalker>
It's something of a joke from one of my classes but essentially it's a negative currency.
<redlizard>
Ah.
<redlizard>
The anonymous votes... is something I know very little about. I know there are protocols for anonymous voting, that can likely work as a base for that point-consensus system above.
<AniSkywalker>
Also, I do want to learn about p2p and blockchains.
<redlizard>
But beyond that, you're pretty much on your own.
<AniSkywalker>
Eh, it's a soft requirement, since I can just not give out names.
<redlizard>
I think it is possible.
MikeFair has joined #ipfs
aquentson1 has joined #ipfs
<redlizard>
So the 30 day timeout protocol as I envision it works as follows:
<redlizard>
- Whenever a peer wants to cast a vote, they publish a signed statement saying "I cast this vote".
<redlizard>
- All peers record these statements along with the time at which they were proclaimed.
<redlizard>
- Whenever a peer decides that >=N votes were cast within the last 30 days, they publish a signed statement saying "at some unspecified point in the past, votes <collection of signatures> were alive".
<redlizard>
- Whenever there are >=M such liveness signatures that each include the same subset S of signatures, that collection of >=M liveness signatures forms a point certificate.
aquentson has quit [Ping timeout: 260 seconds]
<redlizard>
This gives you point certificates that peers can show to each other and verify independently.
<redlizard>
Hm, this version does not even need that onlineness property at all :)
pfrazee has quit [Read error: Connection reset by peer]
pfrazee_ has joined #ipfs
<redlizard>
Peers publish new liveness signatures whenever the set of live votes, according to that peer, changes.
<AniSkywalker>
redlizard neat. That would basically allow me my active requirement too since a peer can decide whether or not a vote is active.
<redlizard>
"active" as in "not superceded by a more recent different vote"?
<AniSkywalker>
So now all that's left is figuring our who's a peer, which is just another log operation--"I invited [pubkey]"
<AniSkywalker>
Yeah
<AniSkywalker>
redlizard since it's just a state calculation
<AniSkywalker>
no reason I can't add more logic
<redlizard>
Yeah, "I invited <pubkey>" should work.
<AniSkywalker>
I actually really like this model. Do you think I need a blockchain to store the top-level proofs and calculate points?
<SchrodingersScat>
so even if we can get ipfs on mars, it'll still take quite a long time to log into facebook, because that can't be public and therefore cached?
<redlizard>
AniSkywalker: I don't think a blockchain is relevant here at all.
<redlizard>
It's just cryptographic proof objects all the way down.
<AniSkywalker>
SchrodingersScat well depends on if we figure out a bookface.
<redlizard>
You might be interested in a blockchain for the (orthogonal) consensus algorithm though.
<AniSkywalker>
redlizard So how do I prove "this proof was valid given the available peer set" or is that a part of the "This is valid" log entry?
<AniSkywalker>
redlizard for the consensus I was thinking Raft
<AniSkywalker>
since config changes are few and far between
<AniSkywalker>
Is SchrodingersScat a real person or a bot?
<redlizard>
AniSkywalker: "proof" is very ambiguous here, please rephrase.
<redlizard>
In my world, there are three data structures involved here:
<redlizard>
There is the Vote, which is a signed message.
<redlizard>
There is the Liveness Certificate, which is a signed message containing [references to] a set of Votes.
<AniSkywalker>
Is that the certificate a threshold was reached?
<redlizard>
And there is the Point Certificate, which is a collection of M Liveness Certificates, such that: there is a set S of Votes, such that each Liveness Certificate contains all of S.
matoro has joined #ipfs
<redlizard>
(Maybe the Point Certificate explicitly contains that set S. Whatever.)
<redlizard>
Oh, and S contains at least N Votes, of course.
<AniSkywalker>
redlizard What I'm trying to prevent is forcing peers to go back and issue log statements for votes that are over.
<SchrodingersScat>
AniSkywalker: I'm all man, baby.
<AniSkywalker>
Not until observed SchrodingersScat
<SchrodingersScat>
be careful what you wish for ;) ;) ;)
<AniSkywalker>
redlizard but N needs to be at least, say, 2/3 of the peers that existed when it was confirmed.
<redlizard>
Hm... that's trickier.
<redlizard>
The above assumed that N is static.
<AniSkywalker>
redlizard I guess I could have peers go back and fill in... but that becomes a blockchain :P
<redlizard>
If N depends on context, things are harder.
<redlizard>
Yeah, then a blockchain may be what you want.
iamtheric has joined #ipfs
<AniSkywalker>
Or I guess I can assume that if a point cert was issued a consensus was achieved.
<redlizard>
No, you can't.
<redlizard>
That is, only between those M people.
<AniSkywalker>
Oh I see.
<redlizard>
M, too, is static in this model.
<redlizard>
If you want context-dependent values of N and M, you need something more elaborate, I think.
<AniSkywalker>
Oh, what if the point contains a list of peers that existed when it was created?
<AniSkywalker>
So it could be verified?
<redlizard>
That works for variable N but not variable M.
<redlizard>
If you want to cryptographically check point certificates relative to the state of the network, then you need a cryptographic certificate of the state of the network, one way or another.
ShalokShalom has quit [Read error: Connection reset by peer]
Caterpillar has joined #ipfs
ylp has joined #ipfs
ShalokShalom has joined #ipfs
ShalokShalom_ has quit [Ping timeout: 240 seconds]
G-Ray_ has joined #ipfs
flapjack0811 has quit [Quit: Leaving]
LouisJencka[m] is now known as LouisJencka[m]1
LouisJencka[m]1 is now known as LouisJencka[m]2
LouisJencka[m]2 has left #ipfs ["Kicked by @appservice-irc:matrix.org"]
AndChat|464025 has quit [Quit: Bye]
nocylah has joined #ipfs
wkennington has quit [Quit: Leaving]
vapid has joined #ipfs
maxlath has joined #ipfs
robattila256 has quit [Ping timeout: 256 seconds]
<ebel>
Has anyone made an init (or whatever) script for linux/debian/ubuntu for ipfs daemon? I could just run that command in a screen session, but there is bound to be something more robust?
steefmin has quit [Ping timeout: 260 seconds]
steefmin has joined #ipfs
kthnnlg has joined #ipfs
aquentson has quit [Ping timeout: 260 seconds]
chaosdav has joined #ipfs
Mizzu has joined #ipfs
<kthnnlg>
Hi All, using new ipfs v0.4.4.5, I pinned a small folder (hash QmUvGoyv8hBprTqjFnhD5m4HGkcxqS4FoNteKEbYmyLj9n) and am sharing the folder from my machine, which running ipfs daemon now. However, for some reason, my other machine cannot get this folder (via `ipfs get ...`). Does anyone have any advice w.r.t. how I find the point of failure?
<kthnnlg>
Btw, the command `ipfs get QmUvGoyv8hBprTqjFnhD5m4HGkcxqS4FoNteKEbYmyLj9n` is what I'm running, and the issue is that the command just hangs indefinitely
cemerick has joined #ipfs
maciejh has joined #ipfs
Mizzu has quit [Ping timeout: 240 seconds]
<ebel>
kthnnlg: I have no idea. for me that same command is hanging
<ebel>
kthnnlg: i have no idea how to debug that
<kthnnlg>
is there some reference hash code that i can use to determine whether i can receive any file via the network?
bastianilso has quit [Quit: bastianilso]
<kthnnlg>
hmm, nevermind... after 30 minutes of waiting, the file suddenly downloads
<kthnnlg>
has anyone else seen this behavior?
<kthnnlg>
it's the second time now
<kthnnlg>
i've seen it
<kthnnlg>
is it usual for a new file to take tens of minutes to become accessible?
<ebel>
There's that readme file that you can download.
bastianilso has joined #ipfs
<victorbjelkholm>
kthnnlg: you can see how many nodes can provide some content by using "ipfs dht findprovs :hash"
<kthnnlg>
victorbjelkholm: super, will try. thx
<victorbjelkholm>
it can take a while to find the file if your nodes are not connected yet. But if you've been running the daemon for a while and they are connected, it should be instant-ish
gde33 has quit [Remote host closed the connection]
s_kunk has joined #ipfs
ygrek has quit [Ping timeout: 260 seconds]
rawtaz has joined #ipfs
rawtaz has quit [Client Quit]
rawtaz has joined #ipfs
gde33 has joined #ipfs
rawtaz has quit [Client Quit]
<ebel>
victorbjelkholm: For kthnnlg's hash, I get 6 results for dht findprovs, returns pretty quickly. but doing a get on that hash is (essentially) hanging...
Hory has joined #ipfs
Foxcool has joined #ipfs
Encrypt has joined #ipfs
bastianilso has quit [Quit: bastianilso]
tmg has joined #ipfs
Foxcool has quit [Read error: Connection reset by peer]
bastianilso has joined #ipfs
<ebel>
ah now kthnnlg's hash works for me. Downloads quickly. Weird.
Foxcool has joined #ipfs
deetwelve has quit [Ping timeout: 264 seconds]
Encrypt has quit [Quit: Quit]
ShalokShalom has quit [Remote host closed the connection]
A124 has quit [Ping timeout: 276 seconds]
A124 has joined #ipfs
jkilpatr has quit [Ping timeout: 258 seconds]
__uguu__ has joined #ipfs
cyanobacteria has quit [Ping timeout: 240 seconds]
Foxcool has quit [Read error: Connection reset by peer]
suttonwilliamd__ is now known as suttonwilliamd
jkilpatr has joined #ipfs
Guest154943[m] has joined #ipfs
<kthnnlg>
ebel: thanks for checking. i was seeing exactly this kind of behavior
<ebel>
kthnnlg: I'm curious of what caused it. it seems... sub-optimal
<kthnnlg>
perhaps i can try committing a new file?
jkilpatr has quit [Ping timeout: 240 seconds]
bastianilso has quit [Quit: bastianilso]
jkilpatr has joined #ipfs
kulelu88 has joined #ipfs
chriscool has quit [Ping timeout: 260 seconds]
webdev007 has joined #ipfs
shovel_boss has left #ipfs ["Leaving"]
dimitarvp has joined #ipfs
slumberproof has quit [K-Lined]
bastianilso has joined #ipfs
suttonwilliamd has quit [Ping timeout: 240 seconds]
Brender has joined #ipfs
suttonwilliamd has joined #ipfs
kulelu88 has quit [Ping timeout: 260 seconds]
dimitarvp has quit [Read error: Connection reset by peer]
Foxcool has joined #ipfs
Foxcool has quit [Read error: Connection reset by peer]
Foxcool has joined #ipfs
santamanno has joined #ipfs
aquentson has joined #ipfs
aquentson1 has joined #ipfs
espadrine_ has joined #ipfs
aquentson has quit [Ping timeout: 240 seconds]
<ebel>
I am running into a similar problem, trying to copy a directory to another machine. And it's just hanging and taking ages
Encrypt has joined #ipfs
pcre has joined #ipfs
<ebel>
oh now it seems to be working. faster. Do i need to keep the ipfs daemon running for a while or something?
pcre has quit [Client Quit]
<MikeFair>
ebel: well requests cahce
<MikeFair>
ebel: When you first request it; it puts it out on its "seeking" list (can't recall the actually name)
<Kubuxu>
wantlist
<MikeFair>
ebel: That propogates as other nodes start seeking it as well and a well defined path
<MikeFair>
ahh thanks Kubuxu
<Kubuxu>
it also tries to activelly find it in the network
<MikeFair>
err and follow a well defined
<MikeFair>
Kubuxu: by making direct connections to the closest XOR node to the address?
<ebel>
I did dht findprovs $HASH, and it returns one node (which is my comptuer). But even after that,it took a while to download
<ebel>
If it knows where to get the file(s) from, why does it take that long?
<MikeFair>
ebel: i've seen that as well; where the stuff I've just uploaded doesn't return right away
<Kubuxu>
my guess is NATs and Firewalls
<Kubuxu>
they always give us problems
<SchrodingersScat>
yeah, seemed like my router wasn't thrilled with something, I run it on a remote machine now until I figure that one out.
<ebel>
hmmm. this is one server on the internet and a desktop behind a nat
<Kubuxu>
Yeah, we opne a lot of connections and fill up conntrackers on smaller routers
<Kubuxu>
it is hard to work around currently
<Kubuxu>
but we are slowly working on alternative transport because of that
<SchrodingersScat>
maybe I can set a raspi3 as a dmz and hook it up with ethernet
<SchrodingersScat>
Kubuxu: thanks for confirming I may not be imagining things on that one.
<ebel>
weird that it takes ages, and then it works like a charm
<ebel>
i can understand it not working at all....
Hory has quit [Ping timeout: 240 seconds]
Brender has quit [Remote host closed the connection]
leeola has joined #ipfs
overproof has joined #ipfs
<grusr>
Yesterday I shared a hash to a folder containing some open movies from the Blender foundation QmS6wQ5842gmBtsfApLKaNwkfmrSYePiJgXm2LRSvQntr2 . If I now want to make a new folder containg these files and some more, is there some clever way of doing that without needing to actually have the files that already is on ipfs as copys on my hdd too? (Kind of a dedup) I have ipfs mounted on my local filesystem if that helps.
<lgierth>
grusr: check out ipfs object patch --help
<grusr>
lgierth: Will do.
betei2 has joined #ipfs
AniSkywalker has joined #ipfs
AniSkywalker has quit [Client Quit]
tmg has quit [Ping timeout: 268 seconds]
Foxcool has quit [Ping timeout: 255 seconds]
ashark has joined #ipfs
Pharyngeal has joined #ipfs
<grusr>
That needs some headscratching before I fully understand it.
<ebel>
hmmm, still not able to get a hash from one desktop to another. Both behind a nat (I think), but I can ipfs ping each one from the other!
<SchrodingersScat>
ebel: charms don't work.
<ebel>
the ping shows that both are properly connected to ipfs network, no?
shizy has joined #ipfs
chriscool has joined #ipfs
<grusr>
lgierth: Nice, that worked. Thank you.
rendar has joined #ipfs
Encrypt has quit [Quit: Quit]
<ebel>
hmm, these 2 desktops are connected by an openvpn network, so there's a local ipv6 network between them
<ebel>
and yet i cannot find a hash from one to the other
<AphelionZ>
morning :)
pfrazee has joined #ipfs
realisation has joined #ipfs
Guest84185 has quit [Changing host]
Guest84185 has joined #ipfs
Guest84185 is now known as amiller
betei2_ has joined #ipfs
betei2_ has quit [Client Quit]
Mizzu has joined #ipfs
betei2 has quit [Remote host closed the connection]
<grusr>
I have updated my folder with movies from the Blender foundations open project, https://www.blender.org/features/projects/ and 24.52GB of material is available at Qmcez6BWo9BrN7iPY1RzydE328UtJRxJGuMUC6uNxYCRy2 , Pin it, get it, watch the mp4;s and webm files from your browser. Hopefully it is of interest.
<Kubuxu>
sorry, it was in the core not in the loop
<kevina>
right, I was going to saw the same thing :)
matoro has quit [Ping timeout: 240 seconds]
maxlath has joined #ipfs
webdev007 has quit [Ping timeout: 255 seconds]
ralphtheninja has quit [Remote host closed the connection]
ralphtheninja has joined #ipfs
OstlerDev has joined #ipfs
cemerick has quit [Ping timeout: 240 seconds]
cemerick has joined #ipfs
mildred1 has joined #ipfs
mildred has quit [Ping timeout: 260 seconds]
palkeo has joined #ipfs
cemerick has quit [Ping timeout: 240 seconds]
webdev007 has joined #ipfs
cemerick has joined #ipfs
pfrazee has quit [Remote host closed the connection]
rendar has quit [Ping timeout: 240 seconds]
pfrazee has joined #ipfs
AniSkywalker has joined #ipfs
<AniSkywalker>
I think Filecoin actually solves the use-case for Climate Mirror / Our Data Our Hands.
<AniSkywalker>
Is it hypothetically possible to have FileCoin without a currency component?
<AniSkywalker>
If we assume peers don't need a monetary incentive to have pieces, but still want to incentivize the rarest pieces and no hoarding, could we use a blockchain to distribute the data and secure against an adversarial model?
chriscool has quit [Ping timeout: 260 seconds]
tmg has joined #ipfs
caiogondim has joined #ipfs
saintromuald has quit [Read error: Connection reset by peer]
gmoro_ has quit [Ping timeout: 260 seconds]
maciejh has joined #ipfs
TheSilencer has quit []
Encrypt has joined #ipfs
maciejh has quit [Remote host closed the connection]
maciejh has joined #ipfs
caiogondim has quit [Quit: caiogondim]
wallacoloo_____ has quit [Quit: wallacoloo_____]
MikeFair has joined #ipfs
<whyrusleeping>
AniSkywalker: the whole point of the currency component is to secure against an adversarial model
<AniSkywalker>
whyrusleeping in what way? I would definitely adapt it so that only certain people can make PUT requests.
bsm117532 has quit [Ping timeout: 264 seconds]
pfrazee has quit [Read error: Connection reset by peer]
pfrazee has joined #ipfs
realisation has joined #ipfs
aquentson has quit [Ping timeout: 240 seconds]
tilgovi has quit [Read error: Connection reset by peer]
aquentson has joined #ipfs
tilgovi has joined #ipfs
aquentson1 has joined #ipfs
tmg has quit [Ping timeout: 260 seconds]
pfrazee has quit [Remote host closed the connection]
maciejh has quit [Remote host closed the connection]
<frood>
if you want to do it without a currency component, don't use a blockchain
<frood>
just use Tahoe or something
<Voker57>
grusr: you might want to use ipfs files
aquentson has quit [Ping timeout: 255 seconds]
tilgovi has quit [Read error: Connection reset by peer]
cyanobacteria has quit [Ping timeout: 240 seconds]
tilgovi has joined #ipfs
<Mateon1>
grusr: For creating folders with a hash already in IPFS, look at `ipfs files --help`. `ipfs files cp /ipfs/hash /path/to/file`. This API is essentially a virtual modifiable folder structure.
<AniSkywalker>
frood Tahoe's not meant for this sort of thing, but we could hypothetically just make the currency for "points" or something to recognize contributions without monetary compensation.
chris613 has joined #ipfs
cemerick has quit [Ping timeout: 240 seconds]
<frood>
you need a really compelling reason to use a blockchain
tmg has joined #ipfs
<AniSkywalker>
frood does "complete adversarial model for distributed storage where peers are shaky and public verification is a must" count as compelling?
nadavh has joined #ipfs
<AniSkywalker>
Tahoe can't handle the instability because it's "active" by nature, shifting data to peers, while we need to allow peers to "accept" data and prove they have it.
romrador has joined #ipfs
<AniSkywalker>
frood could we modify FileCoin so that the coins are created to pay peers, rather than an initial payment by the PUTter?
<frood>
doesn't the coin just function as a reputation metric if you do that?
espadrine has quit [Ping timeout: 268 seconds]
<AniSkywalker>
frood yeah.
<AniSkywalker>
That's basically the point.
<AniSkywalker>
We can see how reputable the peer is (which helps determine stability).
<frood>
is there a reason not to have a central party?
<frood>
folding@home gives out brownie points just fine
<AniSkywalker>
Well, we're dealing with a cluster of nodes that need to be able to operate if we fail (this is federal climate data backup).
<AniSkywalker>
frood the difference is folding@home is folding proteins, we're coordinating file storage where the cluster needs to stay alive (if frozen in terms of new data to store) if the team disappears.
<MikeFair>
o/
<frood>
could do it with federated Storj super-nodes.
<gde33>
when a planet is destroyed!
<AniSkywalker>
hence IPFS :) gde33
<frood>
but yeah, you may be in blockchain territory there
<MikeFair>
What are wondering about ?
<MikeFair>
FileCoin isn't actually going to be a thing for ipfs is it?
<MikeFair>
AniSkywalker: Are you pro-financial payouts to "Farmers" as MaidSafe likes to call them?
<MikeFair>
I've always wanted the computers to keep their own reputations of each other completely independent of the users opinions; the machines would basically manage extended credit and willings to talk to other peers -- that's what the bitSwap methid sounded like (and what attracted me to the project)
<MikeFair>
AniSkywalker: What are you thinking?
<frood>
MikeFair: distributed reputation is haaaaard. and every well-explored model breaks in incentivized systems
<AniSkywalker>
that's the idea behind reputation
<AniSkywalker>
frood except if the incentive is reputation
Encrypt has quit [Quit: Quit]
<AniSkywalker>
which it is for legit peers
<frood>
better phrasing: economically incentivized systems
<MikeFair>
frood: Do ipfs nodes keep bandwidth and latency scores on their connection to peers?
<AniSkywalker>
As a part of the ledger yes
<AniSkywalker>
See Section 3.4.3 "BitSwap Ledger" of the specification MikeFair
<frood>
yeah, BitSwap is cool, but it's strict tit-for-tat
<frood>
doesn't fill the role of say, Eigentrust
<AniSkywalker>
The ledger is swappable.
<MikeFair>
well what's the "threat" ; I always figured it was [D]DOS and data corruption that were the primary concerns
ashark has quit [Ping timeout: 240 seconds]
<frood>
the ledger is swappable, but in an economically incentivized network, you don't know when a peer might profit from sending you a bad ledger
<MikeFair>
frood: I assume that's a short term thing because either you node or someone else's validates the ledger's consistency?
<MikeFair>
but making you do extra work burns the credit with you right
<frood>
validates consistency how? against what objective view of reality?
<MikeFair>
I wasn't sure if you meant "bad" as in "corrupted" (CID address isn't right) or falsified
<MikeFair>
agreement with nearby nodes on what it was
<frood>
gossip with nearby nodes isn't trustworthy when there's money involved
<frood>
and I mean falsified
webdev007 has quit [Ping timeout: 260 seconds]
<frood>
anyway, wrt bitswap, this is all academic. as written, it seems to be purely local.
<MikeFair>
I find a really good model is to think in terms of "energy systems" -- economic incentive -- looked at another way is using energy from the system to "power" something outside the system (aka a drain)
<frood>
that's an interesting metaphor
<MikeFair>
frood: I liked the virtual cluster "upgrade" to the definition of "local" ; equal tit-for-tat across all devices would not work out well (all devices are not created nor used equally)
<MikeFair>
economic incentive is kind of a "transformation" of the energy
<frood>
still relies on gossip though
<MikeFair>
frood: But there's a human organizational component; I also like really random gossip. The argument is that random link changes increase your odds of hitting a bad actor; but they also guarantee that you won't be there long ; perhaps not long enough for a bad actor to do anything meaningful (increasing the difficulty of profiting from the momentary link)
<frood>
that approach only works if most nodes are honest, which is not a reliable assumption when money is involved.
<MikeFair>
oh fwiw, I'm anti-money involvement
<MikeFair>
The system pays the humans to power for its services
<MikeFair>
There's the other problem of "where did the moeny come from" -- what most people don't realize is monetary incentives to pay people is paid for by the future inheritors
<MikeFair>
Everything has to "come from" somewhere
<MikeFair>
The system "provides value" and in exchange, what's valuable to the system is hardware, power, disk space, and bandwidth
<MikeFair>
oh and CPU/RAM
<MikeFair>
I totally see a system being able to give credit to its peers for those resources; and when those peers have exhausted that credit they move to a new peer
<MikeFair>
if tit-for-tat is happening ; the credit imbalance stays minimal
<MikeFair>
but mobile devices convinced me of the need for clustering credit
<MikeFair>
I use my tablet and phone for consumption far more than my SAN/Router/Deskrop
<MikeFair>
s/SAN/NAS
gmoro_ has joined #ipfs
<MikeFair>
my phones due to their construction, contrained bandwidth, and lacking of multiroute peering will never be able to score high in bitSwap
<MikeFair>
but my other devices should
<frood>
whether you're denominating in bandwidth or tokens, if the resource is scarce and in demand, reputation systems break down
<AniSkywalker>
Can I use IPFS & tendermint?
<MikeFair>
So whether it's my devices can send bitSwap around to each other; or my devices can volunteer to offer their credit; something to help the mobiles burn the credit my stationary big resources accumulate
<MikeFair>
frood: but it wouldn't get that way as the heavy user would get blacklisted
<MikeFair>
err users
<MikeFair>
As they started causing a scarcity situation; nodes would effectively "you're too much work to deal with" I'm moving on
<MikeFair>
and then in my mind's eye tell their current neighbors "this peer id is a putz imho"
<MikeFair>
so that by itself doesn't do anything ; putzscores take a while to impact anything
tilgovi has quit [Ping timeout: 240 seconds]
<MikeFair>
but that puts the bad actors as always starting from zero or finding it harder to work
<MikeFair>
But the hard line links are always a factor
<MikeFair>
so giving peers a way to tell the firewalls "I no longer wish to here from [superputz's ip addresses]" help shut them out
<MikeFair>
s/here/hear
<MikeFair>
even if it's just a 24 hour block
tilgovi has joined #ipfs
jkilpatr has quit [Ping timeout: 255 seconds]
<MikeFair>
hey; between that and making CPU resource offers to run distributed algos worth some super-duper swap credit; that's the most I've got on how I see a healthy system could run ; it's not that you can prevent the bad guys; it's detection, containment, expulsion (an immune system that works "after the fact" has been started)
dimitarvp has quit [Quit: Bye]
__uguu__ has quit [Changing host]
__uguu__ has joined #ipfs
AniSkywalker has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]