Non-standard Link Relations in Hypermedia-orientation
A practical guide for future discussion of the gradient of semioticity with respect to semiotic web
A major source of confusion or disagreement within discource around hypermedia web implementations and experiments is its lack of standardization of the link relation. Conceptually link relations can either be bound to a potential or possible action or not: it can exist as an mechanism-independent relation to the context of the resource, or some wider context for which a “continuation” may be necessary to guarantee continuity (or, in other words, consistency in a Service Level Agreement’s implicature toward a workflow as a “service word”).
What this means is that certain actions exposed at a resource (e.g., http://example.com/car/12345) may either set form handling to “wide” possibility space such that a link relation also exposed is not required or present, or “narrow” one in which a link relation must exist for the form handling process to successfully complete its designated task. In terms of the gradient of semioticity, we are concerned with, assuming discovery mechanisms have been put in place, learning as conditioned by of habit-formation, skill-formation, tendency-formation and capacity-formation. An algorithmic semiotics, as opposed to an algebraic semiotics, involves process, adjustment, creation, etc., which is a different mode of determination than translation, deduction, etc., what is fundamentally a mechanistic semantics; i.e., computers are great at mechanically following a step-by-step process themselves, but the consideration of the formal indeterminacy intrinsic to a process which might be named will leave a computer, or set of them, stumped, either giving to the Halting Problem, Off-by-one errors or simple mechanical failure. What we need is a concept of mechanism-independence for a non-mechanistic semantics.
In the context of a distributed system of computation the concept of mechanism-independence implies zero-coordination or coordination minimized determination. Creation, for instance, does not require coordination, in principle, but in a distributed network, typically the problem is solved through consensus (which inherently must account for dissensus), or locking mechanism (see Leslie Lampot’s work on Paxos, which is a commitment-style coordination scheme which uses numerical identity such that higher values are shared for the purposes of effectively determining rank or leadership of a group of nodes, who is “live,” as it were, how structural identity in a formal sense can produce shared understanding in a mechanistic sense). However, the goal of hypermedia-orientation is to fundamentally decouble the mechanistic condition of interaction of client and server, or peer and peer, to the order of allowing cache invalidation to do what it does best: expire resource allocation according to some temporalization scheme (i.e., coordinating time underlying what means what when). Mechanism-independence should also identification-free and scale-free interaction at the given level for any two peers. Hypermedia-orientation allows us to specify, either implicitly or explicitly, through certain patterns of hypermedia factors (H-Factor) made available and expressed in the media type (e.g., HTML, Atom, etc.). But more generally we can think of any text-based expression as capable of bearing hyperlinks, and therefore hypermedia control, which can be paired with some informational content as an event or message which subscends the medium, which is to say: meaning can live in the medium even if it is “vague” as only potential (tendecy or capacity for some node in the network to speak for a unique identifier whether it is a hash or a cryptographic address corresponding to an account such that it is expressed in a unit of that account, however it is mechanistically associated within the network).
A few questions we might ask are, “What is the grammar of link relations? Is it like a dictionary, or telephone book? How can we account for the evolution of names as truth-bearers? Are the link relation names facts? Are they mereological complexes? Are they even names at all? Or are they definite descriptions such that we can give them to processing in First-order Logic, which can be used to determine indeterminate workflows or algorithms? Why should we compose anything in the link relation name at all? If a name were increasingly growing in complexity, would it eventually reveal a structure or will it not?” My hypothesis is that, fundamentally, the question of whether link relations are names or descriptions is moot: there are already link relations with complex structures implicit to their meaning; e.g., prefixing like next-archive and predecessor-version, etc. and version tags in the name itself, such as with P3Pv1, or further camel-cased and other capitalization schemes. We’re already using link relations for temporal relation; elsewhere, I have proposed that we bring into IANA’s purview the research in Segmented Disocurce Representation Theory modeling; e.g., narration, continuation, explanation, elaboration, etc. as link relations which will have a certain structure in their responses that is a pattern different from your typical “web application” semantics as a representation-based form of communication. At one level we can already start to consider how various “modes” of representation-based communication might evolve toward diverent evolutionary paths, and obviously converge: they can possess traits which persevere throughout generations and even reappear as developers (re-)discovery adaptive structure (mechanism-independent structure) that once was successful in perpetuating a medium. Hypermedia-orientation allows web developers to harness the concept of infocological drift: we can measure when hypermedia affordances track to the lowering in frequencies of certain meaning relative to the gradient of semioticity insofar as it is temporalized. If we can think in terms of mechanism-independence, then it seems reasonable to assume that the link relation concept will expand to involve the semiotic derived from the sensible. We might say: nodes in an evolutionary network will less and less treat mere objects but agency; or, web developers will apply the “intentional stance” to APIs (so much for copyrighting them).
Elsewhere in distributed computing systems research that focuses on “program semantics” which deprioritizes “order” or “storage semantics,” the concepts of “apology,” “guess,” and “memory” have been introduced. Generalizing from these concepts, we may tie them back to the concept of a “service word”. In a sense, these terms refer to “processes” which we can also conceive of as (modal) “services.” What could this mean? A “guess” or an “apology” inherently involves some form of contingency: something disorderly has taken place between two agents (nodes, say) in their communication with one another, but fundamentally, a “potential” with respect to failure can be yet another kind of subservice, that which “checks” for inconsistency and does something meaningfully algorithmic about it as part of the shared understanding between the two nodes or agents. They can “agree” on what an apology or guess’s shape or pattern must be before the event of an exchange, and then subsequentially “try again” (retry), “revert” (rollback), etc.; see the Periodic Table of Information. In terms of hypermedia-orientation, what is being suggested here is that the link relation can encode not only service words, but that service words as such can bear a discoverable grammatical structure such that program semantics is associated with an also discoverabe URL within the content-addressable space that is semantic web, or Linked Data. The key question is how it is guaranteed that concept of the necessity of contingency (non-metacomputational orientation, or how to receive modal operation irrespective to ordering or coordination constraints) is given as data with respect to the network structure, for we should never assume that a network, properly understood, is unstructured (regardless of ideological orientations like that of Satoshi Nakamoto, however persuasive they might be). Here we are specifying a conceptual world in which algorithmic medium specificity is made not merely available but accessible even allowing for formal indeterminacy. In a word, meaning happens whether we like it to or not, but we can formally diagnose meaning, so to speak, and what to do about it. However, in order to do this, we must introduce a notion of normative sensibility. The “link relation” can be thought of as a space of possible action, surely, but further, it is a space in which epistemic norms can be given to formal input specification, particularly along the relational schemas implied by epistemic logics (a variant of the modal logic).
Obviously a worry is, as mentioned before, that hypermedia-oriented response data lacks a standardization; in fact, it lacks even the possibility of standardization. Elsewhere I have argued that we can use syntactically trained recurrent nets, a kind of model in Connectionist theory, regarding how to determine when a task is “systematic,” primarily toward the goal on inducing the semblance of cognitive capability (akin to Fodor’s specification of constraint: productivity, systematicity of representation, systematicity of inference, and compositionality of representations). We can think of hypermedia-orientation as “affording” the space of possible action in which these systematic tasks may be carried out within the web conditioned by mechanism and mechanism-independent structure (again, the gradient of semioticity insofar as it can be conceived as a medium specific for meaning to be induced at the conceptual level, assuming stateful interaction actually occurs at the relational level). What the Connectionist model allows us to do is model mechanism-independence insofar as the concepts of capacity, habit, tendency and skill are generally emergent from repetition on a relational structure that allows for syntactic roles to be validated or invalidated according to a distributed history which can be verified, however that means of verification might be implemented. In this case, the methodology by which we fundamentally verify the program semantics is traversal of the shared understanding afforded through targetting the properties of safety, idempotence, mutability and navigability specified in the API’s variable kind of response. If, for instance, idempotent is disrupted, an “apology” can be served to the relevant nodes. If safety is adjusted without fair warning, nodes can “guess again” given some level of constraint before they are to receive an error, or even do something about it (like rollback).
Let’s give an example:
Internally the resource has a particular action which can be invoked or called, but the condition is that it points to a link relation made available to the resource itself. In a sense, its action is bound to an internal stateful context through which its behavior may be afforded. But what does this example mean, particularly? It says:
- ∃xKx There is a process, K
- ∀x∀y((Kx∧Ky)→(x=y)) There is at most one process, K
- ∀x(Kx→Bx) Any node which is a process, K, is in state, B
Naturally the variables are given sufficiently unique names by using a sha256 hash function, which will be stored in a decentralized manner. Moreover, the “process” (K) will also exist as a hash stored at some available storage medium. From a human readability standpoint, the link relation is near-unreadable, except for the symbols representing the actual grammar of the argument. To improve distributivity, the three steps in the argument could be split as shown in revealing its expanded structure. On the client side, the argument is processed following the commonly understood grammar of predicate logic which provides a kind of semantic validation before invoking the href if the user interactions with whatever DOM element which might result in its being called. What’s clear is that the link relation, under certain conditions like with a POST request, will itself not be amenable for use outside of the understood action to which it is bound. Typically, a link relation is only navigational, but in this case it isn’t even navigational: it exists only to satisfy an internal semantic such that an action is executed. So, a link relation’s
[rel] has a “special” name that can be independently processed amongst clients sharing its value. Client side code then can make use of the internal logic of the name such that it uniquely selects an “action” within the actions list. The idea here is that communities of clients are functionally ultimate as to determining how to treat or process exposed link relations despite the fact that the server is providing the actual value. The purpose here is to enable a way of not only caching the determination of meaning in the named link relation, but to provide a way of expressing hooks for automating how that if an ensemble of link relations are exposed, any discovery mechanism can faithfully process the “special” name in such a way that multiple named link relations are invoked in a step-by-step process.
The norms for such behavior, in my experiment here, ought to be expressed in the link relation itself. The gradient of semioticity determines, by this theory, what it looks like to “link data” at the conceptual level in web semiotics, how objects, signs and agents use discovery factors in such a way as to algorithmically arrive at (purposive) meaning in an experimental way (an empirically amenable process). In a sense, I am introducing what in the philosophical literature can be called a “teleonomics,” or a kind of laws of purpose insofar as it is conditioned by normative sensibility (the actions, hypermedia controls, as they are governed by the existence of link relations with respect to a resource insofar as it perceives its own internal mechanism-independent structure). Examples like the XDI Message model gives us a hint as to what a formal structure might look like: link contracts. However, for my purposes, I want to generalize to the possibility of perceptual relations as causal relations: an asthetics for causality given by linkages in the web semantics. I want for the web to be capable of understanding its own harmonization of mechanism and meaning such that causality can b inherent to its structure, allowing for affordance to induce novel modes by which service words express the capacity for causation in a formal yet indeterminant however nevertheless discoverable way.
Perhaps I’ve said enough on the matter for now, or not enough at all. At any rate, I think I’ll leave these notions alone for now and likely take another try some other day.
: https://twobithistory.org/2020/06/28/rest.html and https://doveltech.com/innovation/the-right-end-of-rest/