Pay once store forever

Community
101

This is my cascade collapse hypothesis:

If uploads slows significantly over time, then earnings go down all else being equal … But as earning go down:

  1. nodes are discouraged, so the network counters by giving more tokens …
  2. but this also means the price for storing data goes up … which further suppresses uploading …
  3. which takes us back to number one - a death loop.

So the price of uploading goes higher and higher until uploading stops entirely as none will pay and earnings go to zero.

What cannot go on will not go on … The network thus dies unless we intervene by dumping permanent data.

2 Likes
102

In the end it boils down to the question if there are users… If there are real users data will be uploaded… And if there aren’t nobody needs this network

4 Likes
103

I need it … and I’m sure many people do, but that doesn’t make it economically viable with permanent data.

If we had temp data instead, then it would be just as good as far as I’m concerned - it just would mean I’d need to continuously pay to keep my data on the network. Which is fine for me and I’d expect a lot of folks. The value proposition for me is uncensorability and the ability for others to access that data from anywhere - no central server that can be shut down.

Permanent data would be lovely, but IMO, it’s just not realistic. Now maybe I’m wrong and it is realistic - we are betting people’s data on that hope - and that doesn’t seem right to me.

My cascade collapse hypothesis only has to happen once and it spells doom for permanent data that everyone will have already paid for.

2 Likes
104

I expect there will be loads of demand for this that grows over time.

This might not be a good thing. It would increase supply of tokens at a time when demand is lower, pushing the token price further down, reducing the return on running a node, which may encourage marginal node operators to move on… thankfully it’s a max of 10% of the circulating supply in year 1, and hopefully there won’t be any significant luls in demand after launch so it will be a moot point :slight_smile:

2 Likes
105

We just need to launch a nft/memecoin platform on autonomi and users/upload will be there for sure for the foreseeable future :face_with_monocle:

Or something else completely useless someone comes up with that consumed data

(or something similar to YouTube that actually gets used, an Internet archive thingy or a Wikipedia, a music platform, a 3d model archive alternative to thingiverse…as soon as one app/business is successful on autonomi there will be plenty of users and data… And with the people being there anyway other businesses can strive too… )

3 Likes
106

Just a further thought … At some point someone will fork :ant: and use it for temp data and no permanent data. When that happens, nodes will be incentivized to host that network instead of original because the profitability will be higher -as ratio of recently paid data to old ‘dead’ data is better.

This also leads to the cascade collapse of original network.

I honestly see no way forward here for permanent data past a few years time.

1 Like
107

And I would prefer to store on the original network - even if prices are higher :wink: because it’s not a recurring fee :man_shrugging:

Feel free to move your nodes to the fork with recurring fees :hugs: mine will run on main net

Ps: oh and Blockchain is forever storage too… And nobody cares…(I mean the Blockchain itself)

5 Likes
108

These things may happen rather than ‘will’.

Network effects matter, so forks aren’t guarantee to succeed. Autonomi can add temporary storage if it is clearly advantageous.

If Automomi do add temporary storage, I expect the market for permanent storage will live on, and how popular it is vs temporary will just depend on the relative cost & scale of demand.

Permanent storage can be economically viable while flows of new data are steady or growing, and storage hardware costs per unit keep falling over time. To me, both of these conditions seem likely to persist for a very long time.

But, I think it’s worth considering possible ‘death spiral’ scenarios, as they need to be planned for and avoided where possible… I just don’t see them as inevitable as you do given permanent storage.

5 Likes
109

Sure … that’s true for the sun rising in the morning as well though – read David Hume’s “problem of induction” work.

I don’t see how it can longer term. It’s really all a question of when though, not if. Nothing in nature appears to be permanent.

I just feel it’s wishful thinking now. In the past I didn’t really have my brain around the problem. It’s always bothered me, but didn’t feel confident in a few things - one of which being we didn’t know how the code was going to be written - we had the white paper, but at the end of the day that could have been changed at some point … but it wasn’t and here we are.

After a year or so of allowing this issue to fester in the back of my mind I was sorta itching for an opportunity to present my case. When I did in my first post above I replied directly to David. He hasn’t replied back which indicates to me that he’s either thinking about it or it’s all just too far down the road to change at this time … or both.

Regardless it can, and I say it will, be changed in the future. Further, I don’t see how permanent data can coexist on the network with temporary data long term either, therefore paying for permanent data is honestly just paying for indefinite data – just managing expectations here.

1 Like
110

As I said, it can as long as demand for new data is steady or growing, and storage hardware costs per unit of data are falling.

I guess you see it as inevitable that demand for putting new data on the network will drop off at some point, so my first condition is no longer met?

This scenario certainly a possibility, especially with the likelihood of other competitors emerging. Any network that cannot handle shrinking is vulnerable if demand drops off.

But, wouldn’t that also apply to a network with temporary storage? If a new network out-completed it & nodes left en-masse, whether data is supposed to last the life of the network , 5 years, or 1 year, if a load of nodes leave, surely the network would still die?

I think what is important is that the network has the best chance of continued demand for network services, and continued revenue for nodes so they don’t dissappear… and that will lead to the longest & mosy productive life the network can have.

No, but as others have said, ‘permanent’ here means ‘for the life of the network’, not ‘surviving the heat-death of the universe’.

Permanent markers can’t be used for tatoos, but I don’t think people complain about the category being misleadingly named?

3 Likes
111

Depends on the rate of abandonment. A temp-storage network has higher profitability and more resilience against data loss from cascade collapse as data is always leaving the network anyway.

The point here being that we need to manage expectations. Saying the data is permanent seems like magic marketing and IMO is not true just because people believe it.

Of course … but some things in nature last femto-seconds, others tens or even hundreds of billions of years. I’d prefer if the network could hang in there for a hundred years, not just five.

3 Likes
112

What makes you so certain of this?

We don’t yet know how much people want to pay for permanent storage, and what volume they demand it in… if a good market wants to pay 5x the price for data vs a 1 year contract, it could be highly profitable for nodes, particularly while the network is growing in adoption.

Given the competitive nature of the market for temporary hosting / cloud storage, it may be harder to make a network profitable in that market vs one that’s initially only going to be competing with Arweave.

2 Likes
113

Two reasons:

  1. the cost of storing something indefinitely is a bit of an unknown, so you err on the side of demanding higher earnings. That said, I wonder if average folk running a node from home really take that into consideration, so hard to really say. Longer term though it would need to be true.
  2. New data on a perpetual network also pays for costs of maintaining existing data. So the cost goes up relative to a temporary data network - all else being equal. If hardware, bandwidth, & power go down in price, then this affects both types of network, so we can factor that out.

Our market either way is for uncensorable data. So we aren’t competing with the existing cloud.

2 Likes
114

I don’t see a ‘need’ for nodes storing permanent data being less profitable than nodes storing temporary data. It’s possible that either could be more profitable depending on several variables.

For example, if the network after 5 years stabilised with nodes holding 10% 5+ year old data, and 90% newer data, and uploaders were paying 6x for permanent vs temporary storage, nodes storing permanent data could be more profitable than those storing temporary data. That could last indefinitely if we assumed new data demand was always 20% of network capacity per year. It just depends on the pricing, ratios, and trends.

I’m not guaranteeing that will be the case & I’d need to do some research to get plausible ratios, but given the dynamics at play, it is not inevitable that storing permanent data would be less profitable than storing temporary data (more expensive, yes for sure, but more/less profitable is unknowable at this stage).

115

Actually I agree that there will be a burst. But do not agree that this is a problem.

People are generating a lot more personal data today then they were 5 years ago, mostly more than the last year. 50 years ago in 1974, the university (whole, with some 18,000 students & facility) was storing their data on a 40 & 80MB drive. I take photos that are 30MB in size, so 2 photos swamp that and if I store the RAW & JPG of the photo then less than 2 photos. 3 years ago my photos were 10MB. And people renew phones every 2 or 3 years in general and the phones too have been increasing image sizes years on years. I just did 1TB of videos of my construction of replicas over the last few days. Thats an increase of my whole data store by 2% in one week and expect another 1/2 to 1 TB by end of the week.

So, no, I do not agree that the consumer data drop by a significant degree. All the 10 year olds will be 15 in 5 years and generating a ton of data and larger than their elders did a year or two before. Generally speaking consumer data is increasing at a faster rate than doubling every 5 years, prob more like 4 or 5 times.

A burst yes, but the rate of storage will still be well above die off levels.

And for commercial, the partners are commercial. Also unlike BTC which was crypto, this is storage, not a crypto project. The commercial sector will evaluate and some within a year or 3 (prediction based on experience), be using Autonomi as part of their 3 2 1 backup scheme. From there they may start incorporating it as storage of non-critical (time-access, security) data stores. Many web sites may use it for image storage.

As to covering power, I love how people somehow think using spare resources will somehow cause massive increases in power, etc costs. Incremental drive access is very very cheap power wise. Your computer is already running, it is already using the power. Running nodes may (note “may”) increase power usage enough to register a few dollars a quarter. Leaving your 5Watt led light on for an hour will be more power usage than nodes for days.

If people want to go beyond using spare resources then its on them to decide if its worth it. But that is not part of the design for Autonomi.

Yes we are yet to see if Autonomi will be adopted or not, but “forever” data is not going to be the thing that decides if it dies and in fact may be a significant attractive forces to have people use it.

Also back to the burst issue. That is a more static view where the only people using it will be the ones who are using it now. If Autonomi is adopted (required for anything to survive in internet world) then new people will be using it all the time, so for each person there will be a burst then steady increase after their drop. And that will be happening over the course of years for all the new people doing it. By then all the early ones are uploading much more (images for instance, see above) than what they did 3 years ago.

Your model of what will happen does not account that the users are not static in data production nor the number of users.

6 Likes
116

Punchcard, floppydisk, flashdrive this is the heart beat of storage, yeah I skipped a few heart beats, but storage capacity keeps going up. In all honesty, at the moment I would like to pay 1 nano for 100 TB of data upload to the Network…

Fun for me is the Nebula streaming service uploading their entire 30 TB library, with Google fiber’s 10 Gbps within 7 hours for 1 nano. I watch their over 500 shows and pay 1 nano per GET, not only will they get back their ROI, they’ll have some tokens to upload data again.

117

This is exactly what I was thinking. And this doesn’t account for bugs and energy costs etc that could make the ratio of uploaded data to node storage go negative

1 Like
118

Could a bad actor with LOTS of money simply upload petabytes of junk data to as many nodes as possible until the cost of storage goes so high the entire network collapses as average users simply can’t afford to pay to upload? This is particularly a risk in the early days when there are fewer nodes.

119

In this case there would simply be more people to start up nodes.

Also those that received all those nanos now have extra income to run more nodes too

But yes it would cause headaches and plenty of opportunity for people to earn more running more nodes and the price will drop again

3 Likes
120

Only if people are willing to pay. I think it will become like bitcoin transaction fees. When they go so high people with less money don’t spend whether they like it or not. If this network is for everyday people to use, they can’t be chosing to upload stuff only when the price is low. What do they do when they need to upload but the price is too high? That kind of volatility is not going to work for people who just want to upload their recent holiday photos to the cloud.