Discovering content on the Safe Network cf SafeID and domain names

Apologies a bit of a half baked brain dump but a lot of interesting ideas are springing from something rather simple… and that is the way Rust enums work to voice “what something is”.

So, I could suggest

enum Fruits_I_Eat { Apple::(Braeburn, Bramley), Pair, Banana, RandomOthers }

and that tempts a definition of what RandomOthers is but that a natural way to describe what we are familiar with. Like the best tools, it does not need a manual… even if my syntax is off target it kind of just works.

There is a parallel with what Wikipedia has resolved as topics, titles, and categories.

This had me wonder how people discover content in a Safe Network. Xor is a huge space and there is opportunity to make it more complex than needed. That in turn has led to an exciting idea about what identity can achieve and how to implement that but not to confuse here, I want to first query a few assumptions.

Question 1: Why does Safe Network have SafeID

Surely the SafeID - is seeking to be some proof of private-key but that is a matter for applications handling private-keys and does not need to be the network itself? Obviously, we need from the network the private data function but the access to that and what an identity is, perhaps does not need to be defined by the network itself? The network is base - identity is next level up??

Question 2: What is identity in a Privacy conscious Network

There is a tension about what real world calls for and what people want in certain contexts, relative to notions of absolute privacy. The answer to that tempts different kinds of profiles for an individual users, which are not necessarily limited by number but by type and capability. [Insert my solution for this here!]… but naturally there are options for alsorts of better ideas to arise, if the network is not restricting what identity is…
So, before I progress too far with this, there are a couple of questions that will help.

Question 3: How do users discover content and then each other?

I wonder there is a simple option to see content and other people discoverable, based on declared interests matching and then more-so-even on the basis of where they look for content.

If users look for content based on topic, then surely that is as simple as can be?.. that can be on the back of an accepted standard that follows enum with Parent::Child heirarchy which works by virtue of relating directly to real world object orientated thinking.

I did wonder there’s an option to have an identity declared that is a form of indexing - register your interest and make yourself discoverable but I wonder what is simplest is best and registering your interest, is a hurdle and every hurdle introduced, reduces accessibility and usability. The natural solutions are best, and immediately knowing where to look, is best rather than more complex options for new users at least. So, that lead to another question about NRS.

Question 4: Is NRS too much like the old internet? Should first level domain name NOT be ownable but instead be by topic?

Safe://UK::Politics/[InsertYourDomainName])

The first thought I has was that a general consensus arises, that sees content that wants to be discoverable, positioning itself at a place that is understood. However the normal domain naming in inhibiting what is possible. safe://domain/topic/ limits the option to the host owning that domain but what of everyone spawning everything everywhere?.. what consistency for location can arise? As it stands, the inverse of safe://domain=topic/ will get crowded and the NRS taken immediately the first time it is registered.

So, question - is NRS back to front?.. and what does that mean?

Can we see [Topic] as first level domain, be not owned?.. and then only subdomains are ownable.

Surely that would resolve the domain grabbing problem too?..

Trivially we might wonder how to know under which topic certain content is?.. but does it matter - it can relocate over time, relative to preference… and of course any content can be re-registered and pointers the location as needed from similar misspelled or less preferred topic naming. Consensus of what content there is under certain topics will naturally arise or it’ll be a mixed bag that gives people something to do, resolving that same difference.

What of second level topics?.. do they need a space or just the “first parent topic” will do… or a protocol accepted for this…
I wonder can urls simple accept the enum standard of “Parent::Child” as part of the SafeUrl standard? => safe://Parent::Child/[InsertYourDomainName]/[blog/forum/whatever I own it] … open to extreme as [Parent]::[Child]::DoWeReallyNeedThis/[DomainGoesHere]/ because why not have that flexibility… some topics like Art will have that need for depth?

At the crossing point of topics, that is a location in xorspace, can be alsorts… blogs; wiki pages; forums… and I wonder this tempts an easily discoverable and on topic location for those with certain interests. It makes simpler and obvious what a location is.

It would make applications filtering by topic much simpler too… [more power to the user] principal applies and encourages what is better for useability?

There are unresolved questions that I expect follow… one of which is what is at the top of an empty safe://topic … perhaps safe dog or similar lists domains within that??..

Another perhaps is if topics descriptions are reasonable many - perhaps there are hardlinks available for this - or just the usual bounce forward pointers to the preferred location?

Thoughts?..

1 Like

Just to add to this and not trying to force the thinking at all …

default forum and wiki

When a default forum and wiki template exists, there might be option to put those in place on known topics?

Perhaps we might expect certain default template to exist for each topic… if we can agree what makes a good standard forum for all… something skinable and with option to modify perhaps… but who is owner for that??.. voting etc perhaps allows this in time??

So, we could put a default forum on the normal set of what exists as wikipedia topics, as those appear to have resolved over time for widely people’s interest in those… or if not and for any other, perhaps at first instance of a domain registration a limited set of templates for forum and wiki appear… to save spamming perhaps the cost of first post to topic prohibits spam by being more expensive to cover the cost of posting the default templates for : Wiki; Forum; ?whatelse.

That would restrict registration of safe://topic/forum and wiki but might be useful to consider??

disambiguation

Trivially perhaps there might need to be an option for disambiguation, for rare instances of different meaning - required fork at point of disputed interpretation. That could be just a follow on from default templates handling the top level of each area, instead have pointers to resolve the difference… though a question of who and how disputed is real is confirmed. Voting for certain kinds of organisation … or just accept that the domain space is large enough to cater for two conflicting topics of the same name?

want to avoid the usual echo chambers but some inevitability in needing to cater for what users what and do not want

Applications could whitelist by topic. There’s a risk they see spam or off topic posted to that domain by trolls but some feedback consensus can post to a list of exclude, where that is wanted.
Also, profile type might allow control on the interaction between types of user… thinking alsorts of control is possible with signed proof of type of user; so, trollbots have a hard time irritating real people because they cannot prove themselves and also be interacting with humans.

and with a mind to marketing the concepts of P-S-F

It seems perhaps there are many options possible for actioning Privacy… which is good.

Security comes from the applications used … and with SafeID not being a matter of signing into the network, becomes a splitting off what are Public data and Private data applications, which I wonder is better for user sanity - helps them avoid risk of error?

Freedom then follows from what applications are used for what… and I’m thinking atm users might want different applications by type of profile - certain capability and limits of what is possible will appeal to many use cases.

I’ll be doing a similar brain dump style response, if you’ll forgive me!

SafeIDs are there for the purpose of human readability, memorability, and pronounceability, instead of just a raw public key.

They are entirely optional of course, but the are first and foremost about the usability of human-to-human communication, and recognition.

Just to clarify, privacy, anonymity, while they often overlap, are different things of course.

Identity is a facet of private communication. I can communicate with others and have them know who I am, but so long as I am in control of who sees that communication, and who I am identifiable to, I have privacy.

Anonymity is having my activity visible to others that I choose—or even entirely public if I choose that—but my identity concealed.

So the primary concern, in both cases, is about being in control of my identities, and who can associate my activity with any of them.

They are related again, but discovery of content, and discovery of users are two separate things (that occasionally overlap) that will likely necessitate separate approaches to discovery.

The discovery of users, and how they relate to content, being much more pertinent when it comes to privacy of course.

I’m not sure I really understand this, well, not as a mask use case for content discoverability. Seems more like user discoverability based on social interests?

So the primary use case for NRS is pretty much the same for that of SafeIDs—human readability, pronounceability, typeability over the alternative of XOR addresses.

If we put aside the ownership issue—which is a side topic—there is some merit in the permanence afforded by NRS as it is.

Initial reaction is that this is perhaps quite a blunt and inflexible taxonomy instrument… but remember we have the flexibility of RDF too! So we don’t have to force all this into the domain, when we can have the metadata (which could be much more malleable over time) so it’s thing here.

And perhaps it becomes more socially driven over time too? In the same way pet-names schemes are? And how it might overlap with search too?

Undefined and only as inflexible as domain names.
It’s a simple sort that makes well placed content easy to find… and immediate, rather than requiring someone first build Google. I’ve not seen a workable rdf solution that will empower users to discover, without relying on a third party.

There is of course some inflexibility, if the topic were not to be too large… still, normal websites do seem to centre on topics and content gravitates to what is similar and users push out what is not on topic - try posting politics to What Up Today… even the larger forums necessarily fragment into subtopics, to make those managed and the structure relatable.

In this scheme, who are you thinking sets up these pointers of a specific domain to a category/subcategory? And how are the relevant categories maintained/expanded/governed over time?

No different than now… you buy a NRS and choose what topic that is. The categories follow the interest in them… there is not maintenance… there is no limit on what the topics are - just the syntax would be [parent]::[child] with double colon delimiter.

Even now, the owner of a domain has option to make the next version a pointer to some new location, perhaps even under some new topic. I don’t know the option in Safe Network, if that is network pointer or just forced HTTP header redirect but matters little.

As for misspell or other… if people want to buy domain names to point to their content, then let them do that.

Under topics there would be less likely the domain grab for greed and more likely just an interest in drawing focus on topic. Also, if the top level of TopicThatExists was some dog listing of what content there was, then less need for pointers to catch off target guesses… default to that if no content… so, if you guess at safe://Football::England/forem and that doesn’t exist it lists what else there is under that topic and you find the forum you were perhaps wanting.

There’s no difference to this but for the first element is a topic - and not owned being the significant difference.

The registration of safe://domainname/forum would only work as findable if domainname was meaningful… otherwise the purchase of that, would risk no-one would look for it there. The idea is simply drawing on the interest of the host that people can discover their content - and the users being able to intuitively find what they are looking for… albeit perhaps with a few guesses in a network with little content initially but with a bit of support the idea might be faster to spawn visibility than otherwise.

It’s just channels into which we throw stones rather than a field. The natural language exists because people want to communicate => people want to put public content and others want to find it… and without a trusted third party.

But where and how are the list of parent and child categories maintained and governed?

Or is it literally safe://anystring::anystring/yourdomain?

There is no spoon!

There is zero maintaining and zero governance :open_mouth:

The NRS holds that detail. It is simple and only that the first element is not limiting. So, someone can register safe://English/potatoes and someone else safe://English/carrots.

The topics exist by virtue of there being a NRS domain registration.

yes

So it’s to domains, what hashtags are to twitter…

So if I was infested in a specific topic, I could just ask the client to show me all domains that have been created under safe://fave::specificfave?

That what you are thinking discoverability wise?

That was just an after thought of what might naturally follow… if it’s practical, then perhaps it’s wanted.

I hadn’t made connection with hashtags, as those are typically many and this is one… it’s not intended as a string of hashtags.

I think casting as hashtags would be a mistake, as that would not be guessable… what is guessable is natural language hierarchy at a high level…
so, safe://Houses::Mansions/mycastle perhaps

It’s not exacting - and perhaps the option is still too large… but even then the option for content hosts to come together under one and some fair normal sense of what to expect might work??.. I don’t know how normal people might guess topics; so, a bit of this is relative to the amount of content… with a lot of content it works better?.. with a bit of content, it would make indexing trivial… if we discovered that some exist, others would gravitate to them?.. bit of second guessing the crowd but seemed like an option to consider…