Questions Re: Skynet Model, Now and Future

I have a number of questions about the Skynet model - my goals in asking these questions are:

  1. To be able to update SiaSetup.info with more detailed information on Skynet;
  2. To understand what I would be getting into as a developer if I decided to build a significant project on Skynet;
  3. To help other developers do the same through numbers 1 and 2.

I realize that some of these questions may be somewhat sensitive, but I believe they need to be addressed in order for a serious developer to consider building on Skynet, or for users in general to understand Skynet and make a decision to use it (or subscribe to it) as it matures.

That being said, here are my questions:


  1. Public Skynet portals currently allow content to be uploaded for free. This seems to be the case subject to the utilization of Sia and Skynet being relatively low, as well as Sia storage costs (at least until the last few months). With that in mind:
    1a. Will public portals always allow content to be uploaded for free?
    1b. How long do public portals pin free content for by default today?
    1c. If public portals currently pin free content indefinitely, will this ever change? For example, if storage prices increase above a certain level, if the operation costs of a portal become too expensive, if portals accumulate so many files that a new portal needs to be launched, or some other condition?
    1d. If public portals currently pin free content indefinitely, but this changes in the future, will previously pinned content be pruned? What will determine what is removed if so? How will content owners be notified?
    1e. If public portals don’t currently pin free content indefinitely, do you plan to make it more clear to users that this is the case?
    1f. With the Foundation, there’s a proposal to subsidize all public Skynet portals with Foundation funds to cover portal operation costs. What will be the specific criteria for a portal operator to receive a subsidy? Will they need to run the latest, official portal image or can they customize their own portal? Do you expect that other portals will really be significantly used (outside of those like SkyPortal where the operator does a lot of his own skapp development and other work, and utilizes his own portal accordingly)?

  2. When paid accounts or subscriptions are rolled out for portals, what are the details on how this will work?
    2a. Will users be able to sign up through any portal?
    2b. Will payments be made in fiat, or SC? If fiat, who’s responsible for the payment gateway - Skynet Labs or each portal?
    2c. Will paid accounts be valid only per-portal, or for all of Skynet through any portal?
    2d. How exactly will these accounts be handled? SkyID is being built but is a third-party skapp, and it seems counterintuitive that Skynet billing would be tied to a third-party auth service if Skynet Labs wants to ensure control over the process.
    2e. If there’s a payment split for monthly subscriptions, what portion goes to a portal vs Skynet Labs if someone uses a portal other than SiaSky.net?
    2f. Subscriptions have been mentioned as starting at $5/month - how much content will this get each user in terms of upload, download, and storage?
    2g. How will higher subscription tiers be priced? Will it be a similar flat monthly fee after a certain threshold of bandwidth/storage usage, or will additional use be charged by the *byte?
    2h. How do skapp developers and content creators get paid for their work? Does the $5/month or similar tiers cover these payments, or are payments to creators additional to the monthly subscription? If additional, would these be implemented in microtransactions from an account balance or some other method?
    2i. Considering that a large chunk of a Skynet subscription will go to the above costs to pay hosts and portals, how much does Skynet Labs expect to receive out of the $5/month or other tiers?
    2j. If free uploads and usage are still allowed once paid accounts roll out, will users need an account to use Skynet for free? What will be the free usage limitations? Whether or not accounts are required, how will you keep track of users and prevent them from employing tricks like multiple accounts or use of a VPN to skirt the limits?
    2k. How many subscriptions, or what volume of SC payments at 30%, will Skynet Labs need to see in order to become profitable? What’s the plan to get to that number, and when is it expected?

  3. Developers looking into Skynet are currently being told that the benefits to building on Skynet are that skapps are easy to build and deploy, and that they don’t have to worry about the underlying infrastructure. Some of the above questions pertain to this topic, but in regards to skapps and developers specifically:
    3a. Will developers building skapps on Skynet also be expected to sign up for accounts and pay for the storage and bandwidth their skapp uses? I realize that users using a skapp may end up responsible for some of these costs, like costs for what they upload or download or store through the skapp, but skapp developers may also have large costs - i.e. uploading a complete Wikipedia backup to Skynet.
    3b. If a particular skapp is causing such a load on portals that monthly subscriptions aren’t covering the costs, what will be the course of action in regards to those skapps? Will additional portals be spun up at a loss or through a Foundation subsidy, or will the developer be told they need to cover their own portal costs at that point?
    3c. Skynet development seems to be very by-the-seat-of-the-pants, and SiaSky.net functionality is often modified or broken because changes are pushed to the production server from master before a release. Is this going to continue to be the case? It seems that SiaSky.net should be kept on a stable release so as to not frustrate developers and require them to constantly adapt to unannounced or poorly announced changes.

  4. Regarding the Sia network - relevant to Skynet because Skynet is of course built on Sia:
    4a. Host prices have been increasing because Skynet portals need storage and they pay for it fairly indiscriminately, so hosts can charge more. Are there any plans to attempt to control storage costs for portals, or are those being written off as not a big deal under portal subsidies in the future? If storage costs aren’t a concern for portals, is there a concern about storage costs for normal Sia renters being increased due to Skynet?
    4b. There are a small number of hosts (I believe under 20) which have updated to the latest version and enabled the registry for SkyDB, and only a few hundred hosts and 2 PB of storage on the Sia network in general. What are the plans to recruit additional hosts, and to keep them up to date on the latest releases so that they offer the latest features for Skynet?
    4c. Host improvements were promised after Skynet was launched, but next we saw SkyDB on short notice, and host improvements seem to be on the back burner again. Hosting is technical, unfriendly, and full of issues for many who host, and many of those issues have persisted for years. When can host improvements realistically be expected when there’s so much focus on Skynet lately? I believe that new hosts are unlikely to be onboarded without hosting improvement.
    4d. Is the goal of Sia in terms of hosts still to onboard datacenters with spare storage capacity? If so, when do we expect that this will happen? If not, considering the Sia model of hosts needing to be online 24/7, how many high-quality home hosts do you realistically expect to attract which will fit this criteria and also want to deal in Siacoins and the technicality of being a host? Will this be a sustainable model to build a new Internet?


I realize that this is a lot to cover at once, but if I can understand and answer the above I can better articulate to others why they should build on Sia and Skynet. Thanks for taking the time to answer these questions, and I may have more questions depending on the answers.

3 Likes

First off, let me say thanks for asking these questions in the forum. I really think this is a much better place for in-depth and long term oriented discussion than either Discord or Reddit, and now when people ask similar questions in the future we have a decent place to send them where they can also get involved in the discussion, even potentially weeks or months later.

1a. “Will portals always allow content to be uploaded for free?”

Yes, always. But there is a question of at what level of quality. The sizes of files allowed to be uploaded by free users will be restricted, the total amount of bandwidth allowed to be consumed will be restricted, and other things like number of SkyDB access or HNS lookups may also be restricted.

We plan to add these restrictions all in the form of time delays, meaning any sufficiently patient user can access any non-real time content or application on Skynet while in the free tier. We see the free tier as an extremely important loss-leader to the ecosystem and the place where users will find that there is enough value to justify paying for the service.

1b. Today, portals pin content forever unless the operator manually deletes it or otherwise shuts down nodes. There is not yet robust software in the portal to ensure that content does stay pinned as you add and remove nodes from a Skynet portal.

1c. Yes, right now our formal commitment is to keep files uploaded by free users around for at least 90 days, beyond that we may choose to delete the files at any time based on our ability to retain those files. Cashflow is likely to be the key factor here.

1d. Once we support premium files, we intend to keep files pinned if a premium user has explicitly pinned the file, and we also intend to keep files pinned if a premium user has accessed that file within the last 90 days (365 days for premium+ users).

1e. We intend to open up the conversation significantly once we support premium users. Data uploaded to Skynet is intended to have a strong amount of permanence associated with it, which is one of the reasons we aren’t pruning files yet. We don’t want people to be running around saying ‘data on Skynet disappears’ because that’s not the intended mode of operation for the network. Data is supposed to stick around, and portals are supposed to be able to sustain this with monthly recurring revenue from premium users.

From a branding perspective, it is important to keep control of the narrative here. We do not intend to start pruning files until our premium services are significantly established and users have an easy way to participate in “the permanent Skynet”.

We expect to have the first bits of our premium service rolling out in the next 30 days.

1f. This is definitely something more for the foundation to consider and discuss, but from a growth-of-the-Sia-network perspective, I think that allowing portals to run for free would be a massive benefit as a whole, and would create significant ROI for the ecosystem compared to other initiatives.

2a. Portals will need to add fiat integrations, whether or not a portal can support premium users depends on their own payment structures. We hope to add siacoin support as soon as possible as well.

2b. We hope to support both fiat and SC payments. Each portal is responsible for their own payment gateway. Skynet Labs would love to be in the middle of all of and just send portal operators SC as users sign up, but I think that would potentially put us afoul of money transmitter regulation, and we have no interest in upsetting regulators.

2c. At first, paid accounts will only be valid per-portal. We are doing everything we can to make portals identical to each other in terms of feature set (we still have a little bit more work to go here, especially in terms of standardizing and streamlining releases), so that a user can sign up for the portal that is closest to them and get exactly what they would get for being on siasky.net.

In some ways, we envision portals being like McDonalds restaurants. Whether you are using a portal in China or Japan or UK or US, it’s the same experience. Anyone can pop one up, and if they do a good job of attracting users they can run it at a profit.

We are really hoping to get to the point where if you are paying on any portal, you get premium access to every portal. We have talked a bit about the tech required to make this possible, and SkyDB makes it possible, but we haven’t figured out how to do it easily yet. Our best architecture today would take probably 6-12 months to implement. That said, this is what we also thought about SkyDB and then one morning I was like eureka! and 3 weeks later we had the whole thing running. Maybe portals will get a nice flash of brilliance like this as well.

2d. SkyID is unrelated to premium accounts. It’s also open source, and it’s such a brilliant idea that we may make if a first party app in some form. But, largely it’s unrelated to premium portals or accounts, so not on-topic here. We are currently playing around with MongoDB (traditional infra) to manage premium accounts. Lot more boring, I know :slight_smile:

2e. To be decided. We will probably stick a fee for Skynet Labs in there, after all we do all of the heavy lifting for the portal software. It is incredibly important to us that portals are both easy to operate and also that they are sustainable if they have any significant number of users. Any fees we add will keep this in mind - we want an ecosystem of hundreds or even thousands of independently operated portals, as opposed to most activity happening on siasky.net. That also makes the fee structure important to us, because we want to be a billion dollar company someday and that’s not going to happen if we keep giving things away completely for free.

I will add that we will maintain the open source philosophy otherwise. The code is open source always, and as long as you are willing to pay the fee, you will also get the right to modify and redistribute the software. Free as in freedom, not free as in beer. But that freedom element is super important to us, as long as we’re getting our payment, the code is yours to extend at will. And even you have the freedom add your own layers of fees!

2f. To be decided. Because lightweight users will subsidize power users, we will probably be able to offer significantly more resources than you would get from running your own portal.

We will also support some sort of premium+ or super tier that allows users unlimited access to Skynet, this may have a pay-as-you-go model. Really depends on early UX testing, we want to make sure things are not confusing and are super friendly.

2g. To be decided. But probably there will be 1 or 2 tiers of flat monthly fee and 1 tier that is priced per-usage. Usage would AWS style run the gamut from storage to bandwidth to SkyDB accesses to HNS lookups and updates, etc.

2h. This is a really important question and probably worth being its own thread. Hopefully by the end of the year we will roll out support for app devs and content creators to add a sia address to all of their work and also a price to access / consume / use their work. From the free user / premium user perspective, this will be functionally equivalent to downloading a larger file. Which means if you are a free user, you can still access these files just at a large ratelimit, and as a premium user you can access these files paying right out of your $5/mo.

This payment for creators is viewed by us as the absolute lifeblood of Skynet, and what will drive most of the creators to switch over to creating on Skynet.

2i. Honestly we’re hoping to structure things so that we have really healthy margins. I think the most interesting pieces of Skynet are not the storage, but the content, and content is much cheaper usually than bulk storage. For example, $1/mo in infrastructure could get you 100,000’s of blog post reads. Depending on how Skynet is used, our margins may be quite large.

2j. You will always be able to use Skynet without an account, though creating a free account may unlock a better service tier than having no account at all. To be decided.

2k. As long as Skynet continues growing at even a fraction of the growth rate we’ve seen over the past 3 months, there will be no issues at all for Skynet Labs to be sustainable. Skynet is doing extremely well, and it’s growth over the past 3 months puts it into the top 0.1% of new products. As long as we can keep the pedal down, there will be no issues.

3a. NO! Skynet is paid for by the users and consumers. Except for applications that are many gigabytes in size, developers will always be able to deploy for free and watch their creations grow for free. Any costs will be paid by the users.

3b. We will ratelimit any users that are consuming more than portals can afford. If a single app is responsible for most of the cost, that will manifest in the form of users of that app getting harsher ratelimits until they upgrade to premium. We should be able to structure pricing that this won’t be a big issue in practice.

3c. Skynet development will stabilize soon. There’s a LOT to do to get Skynet to its full potential, and at least while we are still releasing earth-shattering upgrades to Skynet every few months we will continue to be fairly aggressive about updates and releases. As Skynet matures more, we will shift into a more stable release cycle, with siasky.net always being on an official release of the codebase and siasky.dev being home to recent unstable changes.

4a. We intend to address this soon. Skynet portals are price sensitive for storage, they only upload to their favorite selection of hosts. The actual rise in prices is more related to a lot of new educational material coming from the most successful hosts about how to run a host that can charge a lot. I don’t see this as a long term issue, as soon as we are at any sort of critical mass it should be easy enough to prioritize only hosts that are offering the best prices.

4b. We will bang the host recruitment drum as soon as we believe there will be significant reward for the hosts who heed the drum. We have a LOT of hosts on standby, and we don’t want to become the boy who cried wolf lest we lose access to these latent resources. In the event of any sort of situation where the network is actually running out of capacity, we will be able to onboard new hosts very rapidly.

4d. Short term will make use of datacenters with extra racks, long term we expect datacenters to be built out custom for Sia. 95% uptime is super easy, most residential homes in any country with stable power and internet can hit 95% uptime. We believe that there are close to a hundred million terabytes of storage that falls into our “short term” classification, so while custom built out datacenters sounds scary, it’s also not something we believe we will need anytime soon.

3 Likes

Thanks for the answers. My format wasn’t great considering the number of questions, but to summarize what you said:

  • Skynet will always be free to use in some capacity, without accounts, at the cost of rate limiting over a certain threshold for free users or heavy-use skapps in order to encourage users to upgrade.
  • Files are currently pinned indefinitely by default, but in the future a scheme may exist where unaccessed files are pruned after a period of time which varies by subscription.
  • Exact monthly fees and tiers are TBD but the beginnings should be seen in a month or so. They’ll be central accounts initially, managed by a traditional DB. Accounts will be per-portal and each portal will be responsible for the payment process.
  • Content creator monetization will be covered by/included in all tiers, including free.
  • Developers don’t need to pay to upload skapps unless they’re large, which is what I was asking about, so the answer in that case is that yes, developers may need to pay to upload larger skapps with gigabytes of content or data.
  • Skynet Labs is of the opinion that it shouldn’t have any problems being profitable, but no numbers are provided on what that will take.
  • SiaSky.net is going to continue to be on an unpredictable build with rapid changes for the foreseeable future.
  • Portals will eventually try to cost control their storage. There are apparently several hosts in waiting, and datacenters are still the target for hosts. No word on hosting improvements.

If I’m getting any of that wrong, let me know.

Questions based on those answers:

  1. It seems that by subsidizing portals with the Foundation, ultimately this is contributing to Skynet Labs’ profitability if portal infrastructure, storage and access costs are being covered by Foundation subsidies. In this case, basically all of a portal’s premium fee goes to Skynet Labs (or the portal owner) while the Foundation actually covers the infrastructure costs. An arbitrary threshold for profitability before the Foundation cuts out the subsidy has been proposed, such as $5k/month in revenue, but in this case what’s stopping you from spinning up another portal each time one hits $5k/month and keeping 100% of the income while paying nothing for the infrastructure? This model seems to be exploiting the nonprofit organization to cover Skynet Labs’ expenses in operating portals, but then keeping the profits, and I’m not sure that’s great.
  2. What keeps content creators or free tier users from abusing the essentially unlimited funding - only rate limiting? Considering the above, it seems that it’ll primarily be the Foundation footing most of this bill as well.
  3. Can you give any numbers on the profitability expectations? For example, assuming Skynet Labs’ monthly operating costs are $100k, and you get to keep 100% (or 95% after payment processing, we’ll say) of portal subscriptions per the above, you’d need about 21k $5/month portal subscriptions in order to break even. Any solid projections on when this is actually expected to happen? Do you have other numbers that are more accurate?
  4. There was no mention of host improvements - that question was missed entirely. What will be the trigger for onboarding new datacenter hosts? Are any datacenters actually prepared to jump onboard? What capacity are they expected to add?
2 Likes
  1. We intend for portals to be independently profitable. Proposed foundation subsidies are only intended to help get Skynet bootstrapped. Skynet doesn’t need the foundation subsidy for portals but if it doesn’t have one then it’s much more likely that siasky.net is going to the by far dominant portal, because we’re the ones that have a cash fund to operate at a loss until all the profit generating features are in place. Based on current projections, Skynet could be at well over a million users per week by the time that all the profit elements are implemented.

  2. Only ratelimiting. Plenty of services use a freemium model such as this to great success. Should be simple to balance the ratelimits such that the percentage of free users is small enough that the excess from the paid users covers all infra bills. We intend for portals to be profitable independently. If the foundation doesn’t want to help with the bills we’ll have to accept that there will be mostly just siasky.net as a portal until all the profit features are in place.

  3. Based on current usage, a portal would need around 0.1% of its users to upgrade to a premium subscription to cover all of its operating expenses. For siasky.net under its current setup, we estimate it would cost about $1,000/mo to support about 200,000 monthly active users.

  4. Based on surveys and conversations we’ve had in the past we could get more than 20,000 TB online in a relatively short (~30 days) period of time. I am sure that if we needed more, we could be more proactive and get substantially more as well.

1 Like