Tom Banks has been writing papers like this for many years now - I posted about one,
seven years ago.
I would put it in context like this. In string theory - Banks is a string theorist - perturbative string theory (based on the stringy version of Feynman diagrams) provides a model of quantum gravity in flat space. AdS/CFT duality, due to Maldacena and others, provides a model of quantum gravity in negatively curved space. But quantum gravity in positively curved space, such as de Sitter space, was problematic (discussed in
Witten 2001), and became urgently problematic after the apparent acceleration in cosmic expansion was discovered in the 1990s.
Within a few years, Susskind and others came out with a new framework:
"the anthropic landscape of string theory". According to this framework, string theory allows googols of distinct metastable de Sitter vacua, and we live in an eternally inflating universe which keeps budding off instances of all of them.
Tom Banks was an
immediate dissenter against this viewpoint, and together with colleague Willy Fischler, developed an alternative paradigm for quantum gravity in de Sitter space, which they call "Holographic Space-Time". In the landscape paradigm, there is a single physical theory with many de Sitter vacua (with different values of cosmological constant), and space-time can tunnel from one such vacuum to another. In Banks & Fischler's Holographic Space-Time, each possible positive value of the cosmological constant corresponds to a distinct theory.
I don't quite get it, but in HST, space-time is made of overlapping "causal diamonds" (a causal diamond consists of the overlap of a point-in-the-past and its future light-cone, and a point-in-the-future and its past light-cone). Each causal diamond has a holographic description, there are consistency conditions on the overlap of holographic causal diamonds, and each possible HST space-time is a different way to satisfy the consistency conditions, with a corresponding distinct cosmological constant.
The HST framework is inspired by M(atrix) theory, a kind of holography for flat space (Banks is the B in "BFSS"). In M(atrix) theory, everything is described nonlocally in terms of matrix-valued fields, specific localized objects are blocks in the block diagonalization of the matrix, and so on. HST is supposed to work in a similar way. But Banks & Fischler can't exhibit any concrete examples of HST, whereas BFSS was completely concrete. Banks' PhD student Lubos Motl has written some sympathetic but ultimately skeptical assessments of HST, e.g.
2011,
2013.
Closer to the present, the landscape paradigm has suddenly received a more mainstream critique from Cumrun Vafa, who has been promoting some "swampland hypotheses" about what kinds of physics is possible and not possible in string theory. These would seem to rule out the de Sitter vacua of the anthropic landscape. They were originally described in papers known as KKLT and KKLMMT (initials of the authors), but over the years there have been other people skeptical of assumptions that went into these constructions (see: work of Thomas Van Riet). To some extent they relied on the supergravity approximation to string theory and it is possible that they overlooked stringy effects that would completely destabilize them.
So landscape skepticism is a lively topic now, mostly due to Vafa's hypotheses, and this gives Tom Banks's critiques renewed relevance. This may be one of those times when a leading physicist feels that something is wrong with a prevailing assumption, proposes an alternative that no-one else agrees with, but in the long run is nonetheless vindicated regarding their initial doubts. But the question is still, if HST isn't the right way to think about quantum gravity in de Sitter space, what is?