5 ms·

there is a term in use in today's US military - "centralized command, decentralized execution" that mostly captures this sentiment

lifeisstillgood 5y ago
Honestly I am dubious. There is a Norwegian military doctrine that has the mission objectives set and then the "unit" being expected to be autonomous to achieve those objectives. But this was in a time when the Norwegians expected Russia / Germany to take the country and then a Quisling government to issue a recall. The idea was that when a free Norway set objectives they would be carried out even if the capital fell. Does the UMC think that marines should take one set of orders and then ignore any counter-marding orders? It's probably not what that term means. Traditionally command / management has had three goals - co-ordination - resource allocation - decision making in uncertainty The first one is useful, but even in slow moving environments like business it's almost impossible to imagine up then down orders being more effective than horizontal. This is the meaning of your slogan and modern military have learnt this lesson well.(maybe the battle of Jutland ?). It's very dubious any major enterprise has learnt this lesson but I will take suggestions. (Just to expand on this - many business claim to push decisions down to lowest level. As do many militaries. In my experience this is really relegation by negligence - that the upper levels simply do not have the mental bandwidth / time / reporting chains to interfere. For example look at how empires before the age of telegraphy ran themselves - the Satrap was empowered to make wide ranging decisions - but as soon as communication went from weeks to minutes Prime Ministers and Presidents interfered all be time. If your CEO could interfere in your weekly stand up meetings, trust me he would.) This is the essential difference between some new model of organisation (cf Norwegian Mission) and the current one which is just the central office has not automated away your decision making power yet. See Amazon warehouses for an example of how oddly effective that has been (to date) Resource allocation is mostly what we are talking about here. This project gets funding this does not. The failure here was a failure to appreciate the extreme value of anyone doing radio jamming versus that bomber squadron, abs the failure to appreciate the bulge was a counter attack. I am not sure any other system of command (distributed / bottom up / democracy) would do much better. Something tells me a software model of the whole "system" set up and distributed to all decision makers (voters?) would allow a sensible discussion. Not sure it would make a better army. But if it did that army would be terrifying. And decision making in uncertainty is covered above I guess. It is IMO the only truly valuable thing for a hierarchical organisation. And it is so horribly dependant on the taste and capacity of the apex of the organisation. None of these are easy fixes - this is very very much a 20/20 hindsight example - and yet we have to ask does software and the internet open up how we organise large groups of people into new forms. We have run armies and civil services like this for tens of thousands of years. Is it how we should arrange a 21 C business ? Starter for ten: if every body in a SV start up got even shares - that is hiring the fifth person meant your shares went from 25% to 20%. Even founders. Would SV grow fast through automation or through hiring? What other minor changes to incentives will massively chnage how we organise?
- lifeisstillgood 5y ago
  Commenting on my own thought - if your CEO was an AI, then they would be able to drop in on your daily standup and comment on your tickets / powerpoints. which is one end of a spectrum of new org models - the other being (I think) actual independent decision making - something like the difference between a single agent as a company and an eco-system of units as a company. I suspect we need to revisit Roald Theory of the firm
  - selestify 5y ago
    Would the Roald theory not still apply in this case? The AI would drastically lower the in-house transaction costs by decreasing imperfect information by potentially orders of magnitude.
    - onecommentman 5y ago
      “Drastically lower” and “orders of magnitude” are not phrases to be rashly thrown around. AI may hold the promise of faster, cheaper and better, but that remains to be proven by metrics applied to its implementation into existing business processes. Or at least fielded prototypes integrated into existing operations that demonstrate such dramatic improvements. The graveyard of failed IT implementation is strewn with program corpses wrapped in winding sheets/business plans that promised “drastic” improvements.
      - lifeisstillgood 5y ago
        This depends if we think there is a "magic AI to solve it for us" or use of tight feedback loops mediated by software. To overuse an example, logistics is the latter - the software tells each person what box to move to where and when. Decision making is removed. I call this an immutable company - software is released once and that is the company. another release is a new compmay. Now imagine taking this to other organisations. Civil service seems a likely area for much more restricted decision making - can't you imagine how much Congress would like to precisely define what civil servants can do - down to daily activities?

lifeisstillgood 5y ago

Honestly I am dubious. There is a Norwegian military doctrine that has the mission objectives set and then the "unit" being expected to be autonomous to achieve those objectives. But this was in a time when the Norwegians expected Russia / Germany to take the country and then a Quisling government to issue a recall. The idea was that when a free Norway set objectives they would be carried out even if the capital fell. Does the UMC think that marines should take one set of orders and then ignore any counter-marding orders? It's probably not what that term means. Traditionally command / management has had three goals - co-ordination - resource allocation - decision making in uncertainty The first one is useful, but even in slow moving environments like business it's almost impossible to imagine up then down orders being more effective than horizontal. This is the meaning of your slogan and modern military have learnt this lesson well.(maybe the battle of Jutland ?). It's very dubious any major enterprise has learnt this lesson but I will take suggestions. (Just to expand on this - many business claim to push decisions down to lowest level. As do many militaries. In my experience this is really relegation by negligence - that the upper levels simply do not have the mental bandwidth / time / reporting chains to interfere. For example look at how empires before the age of telegraphy ran themselves - the Satrap was empowered to make wide ranging decisions - but as soon as communication went from weeks to minutes Prime Ministers and Presidents interfered all be time. If your CEO could interfere in your weekly stand up meetings, trust me he would.) This is the essential difference between some new model of organisation (cf Norwegian Mission) and the current one which is just the central office has not automated away your decision making power yet. See Amazon warehouses for an example of how oddly effective that has been (to date) Resource allocation is mostly what we are talking about here. This project gets funding this does not. The failure here was a failure to appreciate the extreme value of anyone doing radio jamming versus that bomber squadron, abs the failure to appreciate the bulge was a counter attack. I am not sure any other system of command (distributed / bottom up / democracy) would do much better. Something tells me a software model of the whole "system" set up and distributed to all decision makers (voters?) would allow a sensible discussion. Not sure it would make a better army. But if it did that army would be terrifying. And decision making in uncertainty is covered above I guess. It is IMO the only truly valuable thing for a hierarchical organisation. And it is so horribly dependant on the taste and capacity of the apex of the organisation. None of these are easy fixes - this is very very much a 20/20 hindsight example - and yet we have to ask does software and the internet open up how we organise large groups of people into new forms. We have run armies and civil services like this for tens of thousands of years. Is it how we should arrange a 21 C business ? Starter for ten: if every body in a SV start up got even shares - that is hiring the fifth person meant your shares went from 25% to 20%. Even founders. Would SV grow fast through automation or through hiring? What other minor changes to incentives will massively chnage how we organise?

lifeisstillgood 5y ago
Commenting on my own thought - if your CEO was an AI, then they would be able to drop in on your daily standup and comment on your tickets / powerpoints. which is one end of a spectrum of new org models - the other being (I think) actual independent decision making - something like the difference between a single agent as a company and an eco-system of units as a company. I suspect we need to revisit Roald Theory of the firm
- selestify 5y ago
  Would the Roald theory not still apply in this case? The AI would drastically lower the in-house transaction costs by decreasing imperfect information by potentially orders of magnitude.
  - onecommentman 5y ago
    “Drastically lower” and “orders of magnitude” are not phrases to be rashly thrown around. AI may hold the promise of faster, cheaper and better, but that remains to be proven by metrics applied to its implementation into existing business processes. Or at least fielded prototypes integrated into existing operations that demonstrate such dramatic improvements. The graveyard of failed IT implementation is strewn with program corpses wrapped in winding sheets/business plans that promised “drastic” improvements.
    - lifeisstillgood 5y ago
      This depends if we think there is a "magic AI to solve it for us" or use of tight feedback loops mediated by software. To overuse an example, logistics is the latter - the software tells each person what box to move to where and when. Decision making is removed. I call this an immutable company - software is released once and that is the company. another release is a new compmay. Now imagine taking this to other organisations. Civil service seems a likely area for much more restricted decision making - can't you imagine how much Congress would like to precisely define what civil servants can do - down to daily activities?