Approach
Systems should be able to evolve without becoming opaque.
ABSLTN Labs studies a recurring boundary: computation can propose change, interpretation can learn from it, and people still need a legible basis for what is accepted, acted on, or revised.
Computation can propose.A possible next state is not automatically the accepted one.
Knowledge can remain provisional.Patterns, explanations, and model output can inform attention without silently acquiring authority.
People can revise without erasing.Corrections change what people should rely on now while keeping the source trail and earlier state available to history.
Different people can need different views.A shared state can support participant-relative projections without forking reality into separate copies.
From research to product
Exhibitions make difficult ideas inspectable. Products are kept narrower: they graduate only when a coherent human job can be stated, bounded, tested, and evaluated with real users.