[HN Gopher] What Sets Us Apart: Filecoin's Proof System
___________________________________________________________________
What Sets Us Apart: Filecoin's Proof System
Author : Confiks
Score : 46 points
Date : 2020-03-05 19:34 UTC (3 hours ago)
(HTM) web link (filecoin.io)
(TXT) w3m dump (filecoin.io)
| thulecitizen wrote:
| Filecoin's Proof of Replication sounds similar to Valueflo.ws and
| Holochain's asset backed currencies. Let's go full distributed
| computing and use Mutual Credit while we're at it. Otherwise
| we're just going to carry forward the broken dynamics and
| assumptions of the current economy.
| chrisco255 wrote:
| What is "Mutual Credit"?
| aakilfernandes wrote:
| Whats missing is proof of delivery, and thats the most important
| thing. I'm not just paying AWS to store my files, I'm paying them
| to deliver my files.
|
| _Maybe_ this could work as an alternative to glacier. But even
| then, I doubt it. I 'd take Amazon with an SLA any day of the
| week.
| marknadal wrote:
| Correct, this is what we solved for.
|
| I gave a presentation at Berkeley a few years back, right
| before Bram Cohen (BitTorrent) presented his Chia proof, Proofs
| of Space and Time (before IPFS ripped the name from Bram).
|
| https://www.facebook.com/BlockchainatBerkeley/videos/2006069...
| aakilfernandes wrote:
| Sorry, who is "we"? And got a writeup anywhere?
| marknadal wrote:
| https://web.stanford.edu/~nadal/A-Decentralized-Data-
| Synchro...
|
| https://era.eco/
|
| http://axe.eco/
| ajhurliman wrote:
| That's the other way you get paid in Filecoin, payment channels
| for retrieving files. Although it's not technically "mining"
| it's easier to think about Filecoin having two types of miners:
| Storage and retrieval. Since it's incentivized there's a good
| chance you'll get your files back quickly if you're willing to
| pay more or if it's a common file.
| lalaland1125 wrote:
| Like so many cryptocurrency projects, Filecoin just seems like a
| solution in search of a problem more than anything else. What
| exact problem is Filecoin trying to solve that's not possible
| with current infrastructure? It seems like the decentralization
| adds a lot of overhead and complexity for no benefit.
|
| The main argument that I hear is that somehow Filecoin will be
| cheaper than centralized competitors. However, services such as
| Backblaze are already quite efficient. Why exactly would costs be
| cheaper and why can't Backblaze do those same things and overtake
| Filecoin?
| ajhurliman wrote:
| Storing and serving files a lot cheaper and more reliably than
| current cloud providers seems like something people actually
| want.
| imglorp wrote:
| They needed to compensate providers of storage for IPFS. The
| challenge is how to prove that a provider has your block and
| that they aren't double dipping, ie claiming redundancy when
| they only are storing one copy.
| Ygg2 wrote:
| That's pretty neat. I wonder if similar could be achieved for
| processing power. Distribted hosting ftw.
| Arathorn wrote:
| that's what https://golem.network/ is trying to do.
| momack2 wrote:
| Exactly. IPFS has millions of end users and hundreds of apps
| storing data on the decentralized network, creating demand
| for decentralized _persistence_ of that data. Current
| solution is to either run your own IPFS node or pay a pinning
| service to run a node persisting your IPFS data for you - but
| even better to have a highly resilient marketplace of storage
| nodes that can store redundant copies!
| darkcha0s wrote:
| Where are you getting "millions of end users" from? A quick
| google search doesn't return anything, and that seems a lot
| for a relatively unknown project
| NickBusey wrote:
| IPFS is by no means 'relatively unknown'. It is used in
| dozens of projects. See: https://github.com/ipfs/awesome-
| ipfs
| marknadal wrote:
| I think parent was saying the projects using IPFS are
| relatively unknown, so where are the millions come from?
|
| I have the same question.
|
| In contrast, GUN, has 20M+ downloads/month from known
| sources: Internet Archive, HackerNoon, etc.
| (https://github.com/amark/gun see jsdelivr download
| stats).
| notthegunlobby wrote:
| lol it's not a HN post about a decentralized project
| without marknadal shitting on it and shilling GUN. got a
| running bet it always happens, and you don't disappoint
| <3 -- keep it up!
| johnmarcus wrote:
| Gun is currently a pile of shit. Anyone saying otherwise
| hasn't used it. The sole founder and developer is a very
| nice person, but it's the reality of Gun tech right now.
| 3fe9a03ccd14ca5 wrote:
| I went through that list and can't recognize any of those
| names. Most of them seem clones of "<website> but on
| ipfs".
| msvan wrote:
| Presumably they think decentralization is a benefit in its own
| right.
| programmarchy wrote:
| Technical complexity isn't the only dimension.
|
| Surveillance capitalism and privatized censorship are
| byproducts of data centralization.
|
| Cryptocurrency has a political aspect. Why view it only through
| a lens of technical complexity?
| wmf wrote:
| If a decentralized system is using spare resources that are
| literally free it could theoretically be cheaper than a
| business that pays for hardware. It will be interesting to see
| if that works in practice.
| lalaland1125 wrote:
| One issue is that you are always paying for power. Some back
| of the napkin calculations indicate that Backblaze's prices
| are low enough that it's not worth it to compete unless you
| are getting more than 1 terrabyte per 30 watts as otherwise
| the power costs alone would kill you. That's going to be a
| pretty hard target to hit with "spare free resources" (not to
| mention hardware degradation and capital costs).
| p1necone wrote:
| You're always paying for power regardless of whether you're
| offering your unused drive space on a distributed system.
| klodolph wrote:
| Hard drives consume less power when idle, like CPUs.
| p1necone wrote:
| I imagine that extra power consumption has to be less
| than the combination of non idle HDDs _and_ all of the
| other infrastructure at backblaze, no? And they 're still
| going to be idle most of the time.
|
| And there's the added likelihood that people providing
| hdd space on filecoin are not going to even measure the
| impact of extra power consumption, they're just going to
| notice the revenue. (Not that this is a _good_ thing, if
| someone was losing money, they should know)
| klodolph wrote:
| Agreed.
|
| Cloud storage is a commodity race to the bottom. Cloud
| providers want you to use their storage systems because it
| means that you'll also use their more profitable associated
| services in order to save money on egress. Accumulate
| enough data, and you're locked in because of the high cost
| to migrate. Because IO dominates so many workloads, you
| already get a bunch of bytes for cheap when you are running
| a cloud storage provider. It's already dirt cheap and
| potentially a loss leader (I suspect).
|
| So, I have a hard time seeing how "spare free resources"
| are going to be competitive when you factor in the network,
| capital costs, power, etc. Then there's some unknown legal
| risk that you are e.g. accepting money to distribute child
| porn to people, just like there's a legal risk associated
| with running a Tor node (not commenting on whether this is
| just or not).
|
| Just my opinion. Not speaking on behalf of anyone here.
| DennisP wrote:
| I don't want to start a business that pays a cloud service for
| storage. I want to write an application and set it free on the
| internet to run all by itself.
| johnmarcus wrote:
| Like other cryptos, Filecoins founders don't get rich off
| existing and usually better solutions.
| ThrowAwayGlen wrote:
| I'm very glad I didn't buy into the ICO hype. Filecoin raised
| $257m selling their Filecoin tokens and now, more than 2 years
| later, they still have yet to ship a product.
| paulproteus wrote:
| It seems that Filecoin fundamentally relies on a proof-of-work
| type system called "sealing." The article calls it "an intense
| amount of work" that you must do in order to host a file. Does
| anyone have any estimates or comparisons for the CPU/energy
| consumed?
| whyrusleeping wrote:
| Its currently looking at about 4 hours of single core CPU time,
| plus 2-3 hours more of using many cores for a 32GB sector.
|
| The difference between sealing and an actual proof of work
| based network here is that once you've sealed a sector, you
| don't have to keep computing anything that expensive. If you
| seal 1TB and thats 1% of the network, you should win 1% of the
| blocks (ignoring network growth) without further sealing work
| required. I tend to think about it like a (calculus) derivative
| of PoW.
| paulproteus wrote:
| Cool!
|
| So then, dumb question. Can I delete the file, keep the
| "seal", and keep winning 1% of the blocks?
|
| If yes, then that seems bad for the system.
|
| If not (hopefully), I'd like to also understand what's the
| ongoing compute cost of using the seal plus file together.
|
| I understood from the article that sealing is something you
| do if you are publishing a mirror of a file. Is that right?
|
| Feel free to get bored of this conversation and link me to a
| paper or something. :) I'm grateful for the reply you gave
| above already.
| whyrusleeping wrote:
| Nope, thats where the second Proof comes in. PoSt (Proof of
| SpaceTime) periodically challenges miners to verify that
| they have the data they claim to. Since replicas are
| expensive (slow) to compute, and the challenge window is
| short, this verifies that the miners in question have not
| deleted their sealed replicas.
|
| The ongoing compute cost for these PoSts are quite small, I
| think the current numbers (obviously subject to change) are
| that you have to do one PoSt per day, which is a single
| 100M constraint zkSNARK proof, which should take about 10
| minutes of machine-with-decent-GPU time to compute.
| paulproteus wrote:
| So I guess for 32GB, it's:
|
| - Sealing: Done once. 3-4h compute on one CPU initially,
| plus 2-3 hours on a few CPUs.
|
| - Proof of space time: 10 minutes of GPU time per day.
|
| What are the latency expectations within the system when
| my node is asked for a Proof of SpaceTime? Is it feasible
| to keep the sealing data, delete the file itself, then
| spend ~5 minutes downloading the 32GB on my 1 gigabit/sec
| home Internet connection when someone asks me to do the
| Proof of SpaceTime, and return the Proof of SpaceTime in
| 15 minutes (5 min to download, 10 min to compute)?
|
| Thanks again for engaging with me here!
| chrisco255 wrote:
| What's the purpose of sealing?
| ajhurliman wrote:
| If I'm understanding it correctly, it's to ensure that
| you've actually set aside that disk space for the Filecoin
| network (instead of using it yourself).
| whyrusleeping wrote:
| Sealing ensures that miners have a unique replica of a
| given piece of data on their disk. Since we're basing block
| producer power on storage, we need to protect against a
| number of attacks such as people generating the data on the
| fly, or deduplicating data on disk (claiming to have 100TB
| but only having a single 1TB file 100 times).
___________________________________________________________________
(page generated 2020-03-05 23:00 UTC)