will do, getting closer I think. 
5 Likes
It is also possible to bookmark posts, just wanted to highlight that option.
1 Like
Question: What happens if the the data stored on the network is more than the storage available , say for example there was a mass exodus of nodes, or maybe a mass of nodes got ignored due to a bug that saw them as misbehaving when they were not? Data would have to disappear forever right?
3 Likes
It depends, there are a few things.
So enough nodes die, so does data. Folk stop using network etc.
But for transient issues, nodes can and do hold more than their max records. They can and would republish these on parts of the network collapsing.
8 Likes
Wow ⌠just came back to this thread and saw the explosion of posts. Well since it has everyoneâs attention Iâll just drop the bomb, shall I 
Permanent data wonât work.
Philosophically we can just go with the nature model. Nothing in nature is permanent. There are many reasons for this and some are obvious so wonât list, but if you think about it deeply for a while Iâm sure youâll see the truth of it.
More specifically though, growth isnât constant. Growth happens in bursts with pauses in between. This is true for data. This means that initially we can expect a lot of data to flow into the network, but then it will slow down.
Some think that âscienceâ will provide the big data needed to keep the growth rate up, but itâs not going to. There is no reason to dump raw data on the network. Not many people need to share such data and storing it long term has cheaper solutions.
So what will happen is that the network will grow, then it will pause and many will stop earning enough and shut down their nodes. The price paid (and so charged) for new data then starts to go higher and fewer people will choose to upload ⌠this causes a cascade and the whole of the network mostly collapses with massive data loss.
The solution is to abandon the idea of permanent data and put a time limit on it. This adds some complexity, but itâs doable.
If itâs not done now, it will be done later. Mark my words.
2 Likes
Youâre wrong my friend 
and Iâm sure you already know yourself where to look for the data that contradicts you because it has been discussed repeatedly in the last decade on his forum =D
Looking forward to see the permanent web finally launching
6 Likes
Why would you need science to take this roll when social media is probably capable all by itself?
When will people stop making videos / music / images to share? I think itâs clear that the answer is not for a very very long time.
There will be a price at which the demand for uploading new data to the network will match the incentive needed to make it worthwhile for new nodes to join to store that data. Itâs just a market.
Seems surprising you coming to this conclusion after many years if pondering network economics & not seeing the issue before. What changed? Or have you always thought this & I missed it?
6 Likes
Also if you look at infinite time then yes permanent data wonât work.
What you need to do is define your time frames and remove permanent and replace it with whatâs used and that is forever data in a casual sense of the life of the network.
For instance I can say I am forever getting older. But but is in the sense that its for my life time.
The network will not last forever, something hopefully will be replacing it.
Apart from some global event I cannot see how science will slow down with data creation, or that social media will slow down. For both we are at the very beginnings of the âSâ curve of growth and the prospect of a 1000âs of years âSâ curve of knowledge and data growth in science alone. Imagining sizes is still growing, the quantity of imaging is growing and these too are in the early stages, soon people will be recording 24x7 with 720p or 1080p, then in a couple of years 4K then 8K then 3D 8K and that will have a decades growth path as it goes from a very few to 1% of the population growing to 2%, 5%, 10% ⌠over a decade or so.
Simple communications between people is forever (in the sense of people living)
tl;dr
Obviously Permanent in the absolute sense will see it fail. So will the universe too.
Forever in terms of a lifetime and looking for a replacement network to take over then definitely it has a very high probabilistic chance of working just fine. This is a probabilistic network as David often says, not an absolute truth network. Just the chance of failure/loss/etc is extremely small once the network is adopted.
The last 60/100 years of computing is just the crawling stage of the information age and took over from the invention of the printing press.
Physical paper books are not permanent either are they but look at what is replacing the book age, something greater, so yes the network is not permanent but is very most likely forever for the life of the network till something better replaces it in a few decades
7 Likes
My thinking changed. Itâs been niggling me for a year or so. Iâve dropped quite a few hints in other posts the past year about my thinking here. I think I expressed above exactly why it wonât work as is.
Sure - and markets do fail - all the time in fact, if allowed to. Failing systems are filtered out of a free market economy, this is natural. What lives, what exists, is what remains.
Rather than being too negative here about it though, let me propose a rather simple solution:
When uploading a chunk the uploader appends a future timestamp on the chunk (post-self-encryption). When the price is negotiated, the timestamp is confirmed and agreed to by both parties. Nodes can then see this timestamp on the chunks they pass around; when the current time is past the timestamp, then a node can simply refuse to take and pass the chunk. There may need to be some other housekeeping, but thatâs my idea.
My thinking here is to manage expectations. Few expect books to last forever. But we are marketing the network as having forever data. In truth we donât know how long it might persist.
3 Likes
Youâve expressed that you donât think it will work, but many others, me included, disagree woth the reasoning.
But irrespective of our opinions, we donât know for sure before it exists and is launched.
Given this, I agree it is a good idea to have plans for temporary data that can be added in the future, either if it becomes necessary, or just because it might be an attractive and useful feature for where people only want data to exist temporarily on the network, e.g. for temporary messaging / streaming / or just cheaper temporary storage.
If it does get added at some point, I think itâll be important to ensure it doesnât mess up the âpermawebâ, so e.g. it needs to be impossible to accidentally make links to temporary data etc.
As it stands, we wonât know for sure until it starts to fail, which could be many months or many years ⌠the longer it goes without failing the worse it will be for all those who depend upon it -if and when it does. since we canât know until that event, itâs best to consider the possibilities and do what we can to reduce or eliminate the threat ⌠that means getting out of the permanent data business IMO.
100%.
Perhaps permanent links can be meta-links that redirect to current data.
If/when we get temp data, I think new permanent data may stop. Node operators could simply reject storing it, and would perhaps be incentivized to do so. Which means we would be better off getting rid of the idea sooner rather than later.
2 Likes
I think that the likely long-term cost of storing data on Autonomi will become clear from the trending cost of storage data in the early months and years of the networkâs growth, based on permanent storage.
If it starts to become clear that permanent data alone will price out Autonomi from a bunch of use cases that temporary storage would allow, temporary options could be developed / integrated at that point. I donât think it would be a catastrophic âeventâ, but that over time itâll become clear whether change is needed or not.
While I agree itâs good to be prepared for various outcomes, I certainly donât see any need for Autonomi to âget out of the permanent data businessâ before itâs even been tried!
4 Likes
But we donât need to try square wheels! 
I have pushed for this, and even at one stage it was to be included. Rightly as private data the owner writes as temp data, eg editors have temp files and no purpose in those being saved forever.
4 Likes
@moderators Can this thread be moved to Pay once store forever - #82 by tobbetj
where is seems to best belong?
My 2 nanos worth is that itâs not totally clear that storing everything forever for a single one off payment will work out long term. It would seem to depend on growth in data and nodes. That is a unique idea and the economics are untested.
I think the economics will work. If the cost to store is rising more people will run nodes. Trust the market. I donât see why we wouldnât trust the purest example of a market possible that canât be tinkered with externally.
Promising that all data is undeletable and stored forever for a single payment is a unique proposition and will attract users and data to it just because of those aspects. This will increase the chance of the network growing and succeeding.
This is a very broad statement and Iâm not supplying supporting evidence but unique ideas or ones that take an existing concept and put an entirely new spin on it tend to be the ones that flourish and succeed.
It will be interesting to find out!
5 Likes
except for arweave showing that it at least works for a couple of year and no hint it wonât do so long-term 
so data indicates it wonât collapse immediately and if we see something hinting to non-sustainability weâll have enough time to respond to it
4 Likes
Also anyone who takes their data off their old computer of 5 or more years ago and put it onto their new computer and is not concerned for the space it took.
Yes Autonomi will be challenged if data generation in the world not only slows down but slows to a fraction of the previous year. (like 10% or less). OR people stop using Autonomi which spells a bad time for node operators.
But tell me who is predicting the generation of data will slow to a point where node operators cannot earn enough to cover the few dollars a year for electricity.
But tell me who is predicting that the rate of increase of data generation will slow down in the next 100 years
5 Likes
I predict there will be a burst of data in the beginning, then a slowing as the old data from the previous 50 years that will be uploaded is already uploaded. Iâm talking consumer data here, not commercial. Commercial data wonât go onto the network in a major way for 5-10 years as the network will have to prove itself as being stable before a commercial enterprise will trust it (consider how long it has taken for financial institutions to trust BTC).
So IMO, there will be a burst of uploading, followed by a significant reduction for a long period of time. This will be a period where nodes that arenât earning enough to be worth the hassle will be shut off & people will walk away as they have better things to do â not so much to cover the price of the power as we are talking about average folk running home machines here. Every few years people update their OS and/or hardware ⌠do they decide, sure Iâll keep running in the background even though itâs eating resources, or do they say, well it was fun for a while, but I donât need it and hey whatâs that shiny thing over there âŚ
Cascade failure.
Data needs to die so the network can adjust in size as peopleâs interests change. The network was never going to be a gold or bitcoin-like mining operation. Itâs going to survive as a niche tool for those who really appreciate decentralized uncensorable data and assuming a multi-decade data growth model is unrealistic.
Of course it wonât, but that really isnât the issue here.
2 Likes
You state this as facts though its just your opinion
I predict ever growing data for the time we live and a thriving network that benefits a lot from the promise of permanent data storage and no broken links
(and there might be a reason for new coin being emitted to nodes in the beginning on top of the coin that are paid by uploaders⌠So in the first yearâs even if uploads slowed down at some point there would be additional coin being earned by the nodes⌠)
1 Like
1 Like