3 ms·

For large scale organizations, a reorg is an architecture change to accomplish a change in outcome. It's pretty obvious after you've been at a large org (>1000

For large scale organizations, a reorg is an architecture change to accomplish a change in outcome. It's pretty obvious after you've been at a large org (>1000 people as an arbitrary cutoff) for a few years and you start to see how changes in organizational structure change your own ability to work across teams. I program with a computer. The C-suite programs via org changes.

lifeisstillgood 4y ago
Maybe the C suite should use software. I am serious - i see software as a form of literacy and at some point the workers for the worlds largest companies stopped being human and became CPUs and GPUs. Coders became managers. I mean Twitter is trying to become a company of workers (CPUs) and their managers only. You do need a unifying story so that the managers co-ordinate at a high "mission" level. And most of the rest of the co-ordination is done via integration testing. Frankly "managers" as we know them, supervisory over a few people are done for. What will succeed is what google now seems to optimise for - being nice and safe for coders so they can co-operate. Frankly therapists on tap might make for better org design - iron out most of the directional problems with "the story", keep people from infighting by sharing the cash and occasional therapy sessions and you pretty much describe Google till last year.
- User23 4y ago
  A formal management calculus is an intriguing idea. It could allow something like specifying invariants that must hold for any reorganization or deriving the necessary precondition for some desired organizational property to be achieved.
  - lifeisstillgood 4y ago
    I think you can go an awful long way with three things - a single programmatic API wrapping all parts of a company "datastore" import accounts, people, ... print sales.emea.range(2023,2,1) - a simple forward based test suite - so tests that are specified now but should pass at certain dates in future, or after certain release milestones test.endfebruary.totalemeasales> 5M test.startfebruary.totalmarketingemea.increase = 15% This basically is trying to link budgets to actions in a traceable way. it effectively takes project management and wipes it out as it should. you don't need managers to tell people what is to be done, you just measure if it has been done. (admittedly Inthink things also need to measure like "all posters in shops need to be chnaged by date x" - lots and lots of talking to people

lifeisstillgood 4y ago

Maybe the C suite should use software. I am serious - i see software as a form of literacy and at some point the workers for the worlds largest companies stopped being human and became CPUs and GPUs. Coders became managers. I mean Twitter is trying to become a company of workers (CPUs) and their managers only. You do need a unifying story so that the managers co-ordinate at a high "mission" level. And most of the rest of the co-ordination is done via integration testing. Frankly "managers" as we know them, supervisory over a few people are done for. What will succeed is what google now seems to optimise for - being nice and safe for coders so they can co-operate. Frankly therapists on tap might make for better org design - iron out most of the directional problems with "the story", keep people from infighting by sharing the cash and occasional therapy sessions and you pretty much describe Google till last year.

User23 4y ago
A formal management calculus is an intriguing idea. It could allow something like specifying invariants that must hold for any reorganization or deriving the necessary precondition for some desired organizational property to be achieved.
- lifeisstillgood 4y ago
  I think you can go an awful long way with three things - a single programmatic API wrapping all parts of a company "datastore" import accounts, people, ... print sales.emea.range(2023,2,1) - a simple forward based test suite - so tests that are specified now but should pass at certain dates in future, or after certain release milestones test.endfebruary.totalemeasales> 5M test.startfebruary.totalmarketingemea.increase = 15% This basically is trying to link budgets to actions in a traceable way. it effectively takes project management and wipes it out as it should. you don't need managers to tell people what is to be done, you just measure if it has been done. (admittedly Inthink things also need to measure like "all posters in shops need to be chnaged by date x" - lots and lots of talking to people