Thread by @swardley

X : Do you think cloud will decentralise?
Me : Already has - started with regions and AZs and is now rapidly expanding into outpost, greengrass, EKS anwhere.
X : No, I mean many providers.
Me : There are two forms of decentralisation - provision and control (i.e. authorities) ...

... what is happening is decentralisation of provision (i.e. cloud everywhere, at the edge) but under centralised authorities (Amazon, Alibaba, MSFT, Google etc).
X : What about decentralisation of control?
Me : Many companies working to a common standard ...

... that was the dream of OpenStack but they blew that on day one creating a collective prisoner dilemma. They are trying to recorrect this with K8s but the cloud has moved on, higher up the stack ...

... so, the idea of a marketplace of many different companies each having authority is ... wishful thinking by past vendors and clients they've hoodwinked. Maybe in 15-20 years we will get another go ... open source is a long game ...

... more likely is many points of provision i.e. companies offering EKS anywhere on a utility basis but under centralised authority e.g. AWS.

So, yes, the cloud is going to edge, has been for sometime but it'll be AWS / Alibaba / MSFT cloud that is everywhere.

X : Open source as a long game?
Me : When it comes to unseating a commodity, it's not easy. ASF has the long term view and the community players to make this sort of stuff happen. You won't get there with a "quick rush, grab some VC money and profit" approach.

X : I thought you said 10-15 years?
Me : That was back in 2015. A lot of damage has been done to Open Source since then. You can barely move without some new Open Source Foundation dropping on you. A lot of rebuilding is needed -

https://twitter.com/swardley/status/601511221880291328

... it might well be that the Open Source challenge comes from China. There is some good community work happening there, well worth keeping an eye on.

X : Alternatives?
Me : Amazon et al continue their open source approach and it'll all be open sourced anyway. You'll be able to build your entire stack but you'll still use those centralised authorities because whilst it'll be distributed, it'll be "on chip".

X : "On chip"
Me : Yep. EC2 on chip is already happening, in the near future it'll be Lambda on chip.

X : A Lambda chip in your laptop?
Me : I would guess it'll eventually be in your home router, connected via project kuiper and sidewalk to a world wide mesh. You don't get much more distributed than that.

X : An Amazon router for every home with a Lambda chip?
Me : Well, yes ... I'd imagine that some future combination of Echo and Eero would have this covered. But we're not there yet.

X : What about the telcos?
Me : Well, they should have provided a standardised IoT gateway / hub about a decade ago. Not my problem, they created this mess for themselves with all their fighting over traffic shaping and trying to profit from pipes. Tough luck.

X : Tough luck?
Me : Well, for the future generation of executives who take over. The current crop have plenty of time to comfortably retire and claim that "it was all fine when I was in charge" whilst writing a few memoirs on their marvellous leadership etc.

X : Chambers?
Me : Lol. I normally look at a company seven years after the executive has finally left to determine how successful they were. In the case of Cisco, I'll take a look in 2024 and tell you what I think.

X : Seven years?
Me : Yep. Some of my past research indicates that seven years is the right timeframe. Ideally executive compensation (stock, options etc) should vest seven years after they've left. It would certainly give a longer term perspective.

Me : Such an approach would benefit both shareholders and employees. It would also end sweat and acquire games along with a host of other short term measures. The focus would be much more on the legacy created.
X : Does anyone do this?
Me : Not that I'm aware of.

X : But doesn't they create an incentive to leave?
Me : Well, an incentive to leave with a bright future and the right people and plans in place ... yes ... that's the point.
X : Immediate needs?
Me : Loans against future expected earnings. There'll be market contracts for that.

X : So will this happen?
Me : No.
X : Why?
Me : The problem is the shareholders. It's of benefit to long term holders but not to short term holders. The current system suits those looking for that quick return. You'd have to change the market to start thinking about long term.

X : How?
Me : Minimum terms i.e. if you're investing in a company then you'd have to hold the stock for at least a decade. Back to the old ways of long term investing.
X : Will that happen?
Me : Not in my lifetime. Too many financial instruments are built on the short term.

X : And that's bad?
Me : Well, it encourages certain behaviour i.e. when we detect a change is happening (like a shift from product to utility) then we invest in the long term (i.e. the utility) but still invest in the past world as long as it plays sweat and acquire.
X : ?

Me : The old world needs to squeeze assets, purchase equivalent distressed companies, squeeze more and return value until there is nothing left to squeeze in which case it goes over the cliff. We double up by shorting at that point. So ...

... short term return from old players plus chance of shorting plus long term return from utility players. Rinse and repeat.
X : Long term holdings would stop that?
Me : Yep. We'd have to encourage companies to evolve / adapt. Adds more risk to shareholders. Not going to happen.

X : Given modern political and economic history, I’ve learned not to immediately rule out possibilities.
Me : Fair point. Ok, "not going to happen without a fight" ... too many vested interests.