Thanks, Auke and John, for your inputs, which are enlightening.
In my
A Transaction-oriented architecture for enterprise systems - Sheffield Hallam University Research Archive (shu.ac.uk) paper, I refer to Stamper’s Semiotic Ladder thus (pp.73-75):
“…the ‘cost imposing’ layers in [Stamper’s] ladder structure illustrate the areas in which the productivity of computers (‘the technology platform’) benefits information (enterprise)
systems much better than manual systems. It is where computers are much better than humans, hence the success of technologies such as data processing that we cannot imagine being without nowadays. The ‘value producing’ layers are where humans are better than
computers. Classical experiences with Artificial Intelligence have shown how poorly computers perform…” [note: I wrote the article in 2013.]
If you visit those pages through the above link, you’ll notice I used Conceptual Graphs and Formal Concept Analysis to convey the broader dimensions of balancing business transactions,
e.g. Psychological Stimulation (which, as a side note, I incorrectly—or perhaps correctly—refer to as Psychical Stimulation
🤔 ). These broader dimensions arguably resonate with Stamper’s ladder.
Cheers,
Simon
Dr Simon Polovina
Department of Computing, Sheffield Hallam University, UK
From: Auke van Breemen via CG <cg@lists.iccs-conference.org>
Sent: Monday, July 22, 2024 11:02 AM
To: CG <cg@lists.iccs-conference.org>
Subject: [CG] Re: Stamper's semiotic ladder
John, list,
Thanks, this answer offers the possibility to address Simons ladder question.
In my opinion the semiotic ladder didn't get the attention it deserves.
Also for the reason John wrote about:
And by the way, these issues are the primary reason why I am very critical of the claims about LLMs as a foundation for intelligence or cognition. Most of the people who develop them are unaware of (or deliberately ignore) the challenging
problems of mapping any kind of formalism to and from natural languages.
--
The difference in the ladder between the IT layers (exclusively technical, see Hall) and the higher layers (beside the technical we find formal and informal elements, following Hall again) address this problem. The distinction of the antropologist
Hall between technical,formal and informal culture is pointed at by Stamper. This concerns the translation problem.
But not only for this reason. A second challenge for LLMs is to deal with differences in goal orientation. An identical question may be asked for different goals and also the proximate goal of a question is an answer, but the remote goal
remains hidden unless explicated. The social and pragmatic layers of the ladder provide a scaffolding for dealing with those issues.
Interestingly Stamper started with responding to Morris threefold distinction between a syntactical, a semantical and a pragmatical level. Due to his work in IS, he felt obliged to enlarge and refine the list. This resulted in his ladder.
The group around Kecheng Liu in Reading continued Ronalds work, working their way further up the ladder. Eventually resulting in work on the Assessment of pragmatic interoperability for process alignment. But still this is work done from
the perspective of an IT system as an affordance and it only covers horizontal alignment (between similar layers).
In
https://link.springer.com/chapter/10.1007/978-3-319-94541-5_10 (The social layer of Stamper ladder, just scoll down after the link to get to the text.) I axplore the social layer.
Abstract
The social layer of Stampers ladder addresses the problem of responsible agents interacting with each other. It is the layer at which in organizations decisions are made and transformations negotiated. The method we present supports this
human interplay. It combines principles of actualism, ontology charts, the knowledge in Formation process model and the Cynefin framework to gather and combine quantitative data with qualitative data, expressing attitudes and perceptions in meaningful diagrammatic
representations of business processes. The analytic tool Sensemaker can be used to support decision making.
--
At the same time the ladder needs an extension if we want to shift focus drom IS as a system/object, to an interaction of IS systems/ a semiotic definition of information processes. For that we need Peirce's theory of interpretants as I
argue in https://pdfs.semanticscholar.org/6e1e/b1d74909a117c71d89e8c697c382c7ab7137.pdf.
In daily use the ladder offers a possibility to ask for vertical alignment. This implies that we must be specific about what can be rigidly defined and what not. The ontology chart mediates between my vague approach in the social layer
and Stampers S(emantic)AM (the pragmatic /semantic layer , from this we can go to CG. For one part in order to go down the road towards technial realisation. but also to loop back to the social layer. For which we again need Stampers work.
Best,
Auke van Breemen
PS LLMs cannot deal with the dicent or also dynamical interpretant position in the KiF scheme; it is the moment in the interpretation process the specific goals and context/history of the interpreting system comes ino play in getting at
a conclusion or response. see p 93 in
https://www.scitepress.org/papers/2010/32587/32587.pdf