To use a software term — not a metaphor since the new stack is software — this is the first integration test. We’ve switched with zero warning from half-complete unit testing of point solutions, to a full system integration test at max load.
-
Show this thread
-
The new stack was NOT ready. We were still running integration tests in emulation (aka tech futures), and working through unit tests in production systematically, sector by sector, app by app. What’s more, failure is not an option in this integration test. We cannot rollback.
2 replies 3 retweets 46 likesShow this thread -
There’s no rollback. We’re stuck with this deployment of this integration model of this stack for the time being. There’s no significant do-over or refactoring option here. We’ve locked in a path-dependent full deployment of the software-eaten world. Release candidate 1 *is* V1.
2 replies 6 retweets 45 likesShow this thread -
Now for the “crisis” part. Why is this a crisis? The techlash is over. Silicon Valley tech has won. We’ve been handed the game almost by walkover. We should be celebrating, right? Not quite. This is what I call suspiciously lucky. A crisis just happens to hand us the game?

3 replies 1 retweet 43 likesShow this thread -
Two possibilities: 1. There really were deep reserves of serendipity in the 2-stack system and this really is a well-timed and “lucky” (for some values of “lucky”) smooth changeover. 2. The integration test is failing in all sorts of invisible ways. It’s a house of cards.
2 replies 2 retweets 32 likesShow this thread -
I’m betting the answer is 2. There’s going to be a huge pile of crap in the log file of this integration test to sort through. Full rollback is impossible, but we’ll do a lot of local, partial rollbacks/reverts to the old stack while we fix this in production.
1 reply 2 retweets 40 likesShow this thread -
In a way this is the other side of the yin-yang. Pre-covid was the old stack with patches of localized new stack units. The rollback will be new stack with patches of old stack units. It will be an unholy mess that will take a decade to cleanup.
4 replies 3 retweets 51 likesShow this thread -
So if this hypothesis is correct, the reboot problem is a tech problem: how to finish the unfinished stack that has been hot-swapped in? I think if the economic approach aims at this problem, it will “take” otherwise it will not, and trigger a secondary larger technology crisis.
3 replies 1 retweet 22 likesShow this thread -
By which I don’t mean a tech market crisis like the dotcom bust, I mean an actual tech crisis like Y2K. But way harder.
4 replies 1 retweet 20 likesShow this thread -
Venkatesh Rao Retweeted Venkatesh Rao
https://twitter.com/vgr/status/1256100616553721856?s=21 …https://twitter.com/vgr/status/1256100616553721856 …
Venkatesh Rao added,
Venkatesh Rao @vgrPolitical outcomes are also at the end of a supply chain with a pattern of consent of the governed as the raw material, political apparatus for making policy as factories, and executive as supply chain. The political supply chain is as broken as goods, services, and money ones.Show this thread1 reply 0 retweets 1 likeShow this thread
Venkatesh Rao Retweeted Venkatesh Rao
https://twitter.com/vgr/status/1256068731882450944?s=21 …https://twitter.com/vgr/status/1256068731882450944 …
Venkatesh Rao added,
-
-
-
Replying to @calhistorian
Done! We've saved this whole thread to your Readwise library so that you can revisit/remember it

0 replies 0 retweets 0 likes
End of conversation
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.