Our team member @JimCollinson has responded this morning to some of the recent questions received as to how Autonomi is different from other technology. Check out the post on Twitter/X
Here’s a breakdown of the current alternatives:
IPFS: This is the bread and butter of data storage for Web3 at present. Again this is nifty tech. Content addressable data is needed. But it doesn’t store anything. It creates pointers to data. Data that is stored elsewhere (predominantly on traditional servers) So progress, but only part of the puzzle.
Filecoin (FIL): This is a decentralized marketplace for IPFS data storage. You buy coins in a decentralized way… and negotiate with a storage provider to store the data for you in a traditional way. And you have to figure out between you how to get the data to them. Plus, you’ll still have to renegotiate year-on-year, and pay the bills to keep that data alive. Useful marketplace, but not quite fitting our requirements in full yet.
Arweave (AR): Permanent storage of data. Great! But not private, nor encrypted. Boo! Maybe you can roll your own encryption? I don’t know. But why can’t we all just have end-to-end encryption all the time for all the things. Is that too much to ask?
Internet Computer Protocol (ICP): Once again, great tech, and pushing the boundaries. Like all these others, allies and friends. But. Not permanent data. And the barrier to entry is very very high. So high that some might say it has a natural tendency toward centralization. If you want to contribute your resources to the network and run a node, you’d better have a data center and a few $100k to back it up.
So, just imagine if we could combine:
Content-addressable data like IPFS
Decentralized marketplace of Filecoin
Permanent data of Arweave
Have everything encrypted all the time as standard
and have a massively distributed network of millions of everyday devices …
I can’t stress enough but Arweave won’t use encryption because it goes in contrary to their own terms.
They have at least three layers of censorship and a price of what I’ve seen being betweeen $15-25k per TB and I see it really jumped by 1000$ during just today which might be frustrating for users.
Autonomi price is normalised and doesn’t depend that much on token market price volatility rather focusing on storage available.
Looks like there’s a new kid in town, https://www.walrus.xyz/. It seems they have made some waves since their testnet release a few months ago already; plus there are similarities to Autonomi. Like file division in multiple chunks and replication x5.
Biggest difference is it’s blockchain based of course… But from their saying, they rely on the fastest existing blockchain (The token is called SUI, #14 on coinmarketcap), developped by the same team.
What caught my attention is that a dropbox like project relying on Arweave recently made the switch to this walrus tech. The guy posted the reason why here : x.com
They don’t have permanent storage built-in as far I know. But could be worth having a look. Some people even mentioned it in the altcoindaily x autonomi yt video comments.
Their branding is cool, so it seems like a serious contender with a big audience already. I didn’t take the time to test it yet though!
Yep, most staking allows the richest to keep buying more from their staking earnings and staking more and buying more to stake to earn more to buy more to stake and earn more …
Similar effect to compounding interest has on a “bank account”
Yeah someone mentioned it in response to my post here as well: The difference with Autonomi - explained
So I’ve added it to the overview. And with a bit of googling I do think it is quite a good technology.
Considerations:
Its based on a blockchain
Content isn’t automatically encrypted
No perpetual data
There is ongoing storage cost
Running nodes requires quite some powerful hardware
Even though data isn’t stored perpetually they do have an interesting way to store it. You pay beforehand, with a maximum period of 2 years. But you can as well set up a public fund in which people can contribute, so the contract is automatically renewed and that period of 2 years is then extended.
That is one of the largest inherent problems I have with Autonomi. Deleting data isn’t possible, so it will only grow. This principle has huge advantages as well, but the downside is that for short term data Autonomi will probably be more costly.
All blobs stored in Walrus are public and discoverable by all. Therefore you must not use Walrus to store anything that contains secrets or private data without additional measures to protect confidentiality."
The cost of storage on Arweave is currently between $15–$20 per GB. It has gone as high as $40 earlier in 2024. This high upfront cost guarantees storage for 200 years. But for many use cases we don’t need storage guarantees for two centuries, and paying for such a high per GB cost rules out storage at scale. Large data migrations in the 100s of TBs creates a 7-figure upfront cost, which no enterprise will accept in a quarterly budget and no institution will ever come close to affording. Even on a personal level, our most precious files may run to a few terrabytes, that may would end up costing $50k.
This major problem plus the increased possibility of a cascade collapse - nodes leaving as not enough new data to fund old data; are solid reasons why we need an additional pay as you go model on
has many advantages over other competitors, but if we don’t add a solution for term limited data, we cannot be competitive in the end. Permanent data will only ever be a very small market and we are hobbling ourselves before we even step out of the gate.
I don’t think this is the case, as it’s a model that should work well for all kinds of public / published data e.g. video sharing, websites, music publishing etc etc. These alone seem like a pretty big market.
It would be interesting to see a breakdown of why costs on Arweave are so high though… at first glance it does seem extreme: $15 per GB is around 40x the raw cost of hard drive space (assuming $25 per TB), and I wonder if it’s likely to be similar on Autonomi in time?
Maybe the spike to $40 indicates it’s not easy to spin up nodes for Arweave in response to demand?
I guess 40x raw cost may be fair for data on Autonomi; if data is replicated 5x, and then ~8x for no time limit perhaps the market on Autonomi will settle at a similar price.
If this is the case, maybe it’ll be worth working on temporary data post-launch, along with compute, smart contracts, and other things the network could offer to add value, but for now I think the offering is fantastic and they should just crack on and launch it, then take things from there regarding ongoing developments.
It is what people expected from websites only to be disappointed, annoyed, and angered by when their favourite movie, song, documentation, funnies pages, and so on disappear because they expired etc. Basically people have learned to live with it, but really do want it.
Also saying it is such a burden when plenty have demonstrated with figures, experience, etc that it can work, does not make it such a burden.
I can only agree that there will be a market for it … but if it’s a choice of a much cheaper rate for temp data that I can simply renew on a regular basis, then for most business use-cases, they will go for that - for a similar reason they lease instead of own assets.
Also as permanent data isn’t yet a long term proven thing, it could take a long time for it to be accepted at the higher rate.
Nodes storing permanent data too, will likely seek to err on the high side of pricing if they are cautious … and if we end up with a network full of nodes that are not cautious, then we increase the risk of the cascade collapse scenario down the track.
If you have data that you really want to keep and preserve forever - as I do, then you don’t entrust it to something new in the first place … you go with the tried and tested - even if you have to pay renewal fees.
With permanent data we gain some convenience, but also risk, and from the perspective of a professional data manager, that risk is likely going to be far too high with something like or Arweave until they’ve been around for a decade or so … hence I wager we will have low demand for permanent data.
And adding to that I am still not convinced that the economics of the network will work overall. I have yet to see a product on the market where a consumer is paying a finite amount for a service to be provided for eternity (with costs assciated to it).
This will only work it the network grows infinitely with new data uploads basically paying for the (diminishing but never (!) zero) cost of storing old data.
Resembles the economics of a pyramid scheme to me.
There is an assertion here that non-permanent data is much less of a burden to the network. However, there is evidence that this may not be a concern.
For example, is the storage cost the overwhelming factor in price or is it the bandwidth used when accessing it? Are dormant files actually much of an overhead?
Also, we have deduplication, which @neo mentioned may also be for private data too now. If a chunk already exists, then it won’t add another copy - the burden on storage has not increased, where as 2 temporary copies, with different chunks, will be double the burden. If there are 100s, 1000s, of copies, this is a massive net gain.
Note that when we discussed this before, to have chunks unique to a user, they would need salting to be different, to ensure they can be safely deleted.
