When your boss is an algorithm (talk outline)

The Future of Management    - Orders of magnitude lower cost for anything  programmable         - dispatch, metrics-driven evaluation         - already for a lot of things         - still haven't used deep larning or other advanced AI to this or PM, so - _the real tsunami is just about_ <- this         - therefore: collapse of cost of management across the board (and improvements on frontier of possibility)     - good for: orgs with lots of glue (non-domain mgment overhead)     - bad for: glue     - good for: org size --- Amazon, Uber, would be much more expensive to build with less automated ops     - weird: fluid organizational limits         - automatization --- machine-to-machine comm standards --- integration a la logistics         - a new "theory of the firm"     - arguably: a new jump in productivity. possibly.     - pov: employee/"employee"         - sharp differences between algo- and human-led in strategies and work experience             - not necessarily less discrimination (algorithms reflect) but easier to fix with good will or good legislation         - narrow metrics means narrow training; used as FPGA.              - long-term Starbucks effect: not building a lot of human capital         - "millenial disloyalty"? consciously or not, they know they are being led by (primitive) algorithms and metrics.             the only reasonable reactions aren't loyalty (to whom?) or anger (ditto) but disengagement and optimization.                 in other words - algorithmically.     - pov: career managers         - oh boy. won't disappear, but won't be a lot of openings.         - number of career managers ~ number of human employees / efficiency of management. both numbers moving in the "wrong" direction         - climb up quickly, or manage "artists"/"art" (anything you can convince the industry can't be managed by algorithm, because reasons)     - pov: business opportunities         - think about going larger, think about going more complex. both possible.         + op: size control and no status bias: riskier ops w/faster shutdown         - think fluidly: endosymbiosis is possiblenow.         - think laterally: parametrizable orgs can move between domains more easily.         - mgmg can make you a more attractive merger target.         - not metrics - cybernetics. a lot to study and know-how to add. not many MBAs up to speed.             - as usual: new tech collapses demand for average managers, talent shortage for the best ones, the ones who build cybernetic orgs         + op: AIs can use more exogenous resources faster, changing faster (opps for external providers operating on AI time)         - bizseeds, AWS for bizops --- scale up from 0 to 1 to large as easy as IT ops.              - outsourced mgmt (a la upwork, but better and more)         - don't overlook *complexity*; don't limit your model to what you can manage.         - endosymbiosis as a possibility: more porose orgs allow new business models as intra-providers and so on.         + op: domain- or strategy-specific knowware development and selling or renting.         + op: open-ended computers find optimizations not only better, but weird. the best and newest bus opts might simply be the ones AIs will fig out.         + op: delegating power + crypto gives you power and protocols in game theoretical settings --- optimizations possible, and new business opps         + op: large-bandwidth strategy (try a hundred things at the same time, optimally, know your customers one by one)         + op: AI-human society interfaces         + op: algorithmic transparency to make possibly currently discredited or illegal businesses?