[This fragment is available in an audio version.]
This week saw the cancellation of the Australian Stock Exchange’s long-running effort to build a blockchain-based trading system. Which, oddly, has me thinking of 2016, when AWS decided not to make a strategic investment in blockchain, with my input a contributing factor. It felt like a good story while it was happening.
Since I left AWS in 2020, I’ve been super-careful not to share things from behind the scenes. I can’t actually remember the details of the nondisclosure agreement, but I have strong feelings about the ethics. This story, though, doesn’t reflect poorly on anyone and I’m pretty sure nothing in it is material to any business plans at AWS or elsewhere.
Andy meeting · At some point in mid-2016 I got hauled into a conversation with Andy Jassy. I can’t remember if it was video or f2f, can’t remember how many of his staff were there. There were four of us present who were senior techs, not Jassy staff.
Andy is an outstanding communicator and was eloquent on this occasion. You have to understand that one of the most important parts of his job was listening to the CIOs and CTOs of huge enterprises explain their problems and concerns.
He said something like this: “All these leaders are asking me what our blockchain strategy is. They tell me that everyone’s saying it’s the future, the platform that’s going to obsolete everything else. I need to have a good answer for them. I’ll be honest, when they explain why it’s wonderful I just don’t get it. You guys got to go figure it out for us.”
Well, OK then. I can’t remember whether it was right there in the room or by email after a short caucus, we got back to Andy along the lines of “We mostly think it’s mostly bullshit and probably not strategic for AWS, but we’ll look harder.”
Before I move along, Dear Reader: There was a dead give-away in Andy’s presentation of the problem. I’ll get back to it later but do you see it?
The hunt · We went looking around the industry from behind our screens and discovered a few things:
Actual working business applications of blockchain were really, really hard to find.
Plenty of blockchain products were on offer, characterized as “polished”, “robust”, “production-ready”, and “regulator-approved”. But if you looked hard at their customer stories, it got pretty vaporous pretty fast.
The throughput of proof-of-waste blockchains was just as bad as we thought.
In practice, the technology is a database. Everyone who did anything had some sort of a database structure mapped over the actual blockchain, with the usual B-trees and so on. It wasn’t obvious how anything would be different if there were something other than a blockchain behind the B-trees.
The Australian Stock Exchange was betting the farm on blockchain, which seemed to prove this was no joke.
A huge, almost incomprehensible, volume of venture capital was flowing into the sector, and that money was localized in the Finance sector, specifically in Manhattan.
AWS was already making a lot of money off blockchain. All these venture-financed companies had to build out infrastructure, and most of them were all-in on cloud, either AWS or GCP (don’t think Azure got much of that biz). So there was a serious flow of cash from VC firms into AWS.
We found ourselves in conversation with the Finance-sector sales organization, naturally headquartered in Manhattan. We got their entire attention right away, because they’d noticed the crazy venture-money flow, and they were getting questions, the same kind that Andy had described, from their institutional clients.
The right thing to do was obvious.
New York, New York! · In August of 2016, four of us went down there for a few days. We talked to big established Wall-Street firms and a few startups.
There’s this amazing thing about working for AWS. You can get the attention of the IT leadership of pretty well any organization in the world and say “Hey, could we drop by and talk about…” and most times they’re happy to do it.
(That weeping you hear is the former me, the one who had two startups and spent months bashing at the doors of enterprises in the usually-vain hope of getting anyone meaningful to look at our technology.)
So, no doors were closed to us. Which was pretty nice, except for it was stinking hot that August and we were on multiple steamy subway trips every day; got back to our hotel rooms pretty well emptied out.
(The pictures decorating this entry are from that trip.)
The questions · We really only had two questions, both for the big-finance players and for the startups. “What is it you want to do?” and “How does blockchain help?”
The answers, to our disappointment, failed to shatter any preconceptions. The things they wanted to do were perfectly reasonable. Some of them were damn exciting. They all needed databases. They could all make use of ledger-like data structures, also cryptographic hashing and signing. But, um, why did they need blockchain? Severe lack of clarity on that.
The key moment was when we got in a room with the CTO of this one startup, in Tribeca I think. When I heard their VC funding number I thought it was the valuation, not the investment dollars. The customer list was blue fucking ribbon and don’t you forget it. These guys were razor-sharp.
They presented some of the systems they’d built and yep, we were impressed. Then, with the startup CTO in the room, one of my fellow engineers asked the key question: “All these systems, are there any that wouldn’t work without blockchain?” The guy didn’t even hesitate: “No, not really.”
And that was about that.
Obviously we also talked to a few leading lights from the crypto scene. That wasn’t very helpful, because they seemed mostly concerned with the aspects that got you out from under troublesome government regulation and contract law; it all had an unsubtle aroma of libertarianism.
Lots of them mentioned the Australian Stock Exchange and said you must be a peasant if you questioned technology that was about to be adopted in such a Serious Enterprise full of Serious People. After all, it was being built by Accenture, no less! (Cue eye-roll.)
And then there was the other faction, all about number-go-up-we’ll-get-rich; these were the days of ICOs, each sketchier than the last.
There was still a gaping hole when we asked “What useful thing does it do?”
What Andy said · It looked like the conclusion was going to be “no there there”. We shouldn’t have been surprised. Remember Andy Jassy saying “when they explain why it’s wonderful I just don’t get it”?
At this point in history Andy was possibly the world’s single most accomplished person at listening to people talk about Enterprise IT problems and the tools needed to fix them. If he couldn’t see the value in blockchain, that was really a dead give-away right there, at the business level. Maybe looking at the technology side had been a waste of time.
An AWS service? · We didn’t just try to ascertain market realities. We asked ourselves “OK, if we had a distributed ledger technology, what would we build with it?” then came up with a few proposed AWS services.
Mine was a database with special features optimized for representing fungible units of value (you could call them “tokens” but I didn’t) that could could be exchanged between parties, with each such exchange accompanied by a description of whatever was being exchanged for the tokens. I thought the schema setup for describing the “whatever” was pretty clever at the time, but I forget it now.
[Tim, so you designed NFTs? -Ed.] [No comment. -T.]
It included a shared public ledger with crypto-provable integrity that recorded all this stuff. But no blockchain because we just couldn’t convince ourselves that the real world wanted zero-trust; so there was a transaction manager you had to trust.
Dear Reader: I think that at some point, in a civilization, there has to be trust. I think that’s maybe the main reason we have civilizations. Call me crazy.
Do I think AWS should have built this thing? Eh, I can’t get excited either way.
The end · I can’t remember the details of how our findings got back to Andy. They were of the form “Ledgers are useful, cryptography tech is useful, blockchains aren’t, the field is full of grifters, but we could build distributed-ledger infrastructure and then these cool services on top of it.”
I seem to recall that about a year later we got another outreach from Andy’s office saying they were still getting blockchain heat, oh and Australian Stock Exchange. Could we have another look and get back to them? We did and I think the only thing that changed was we said “Keep an eye on Ethereum.”
Subsequently another group decided that if people were spending their VC money building crypto infrastructure on AWS, we should help them do that more efficiently, and thus was born Amazon Managed Blockchain, offering Hyperledger and Ethereum platforms as a managed service, in the typical AWS style.
I wasn’t involved and have no idea how that product did; wouldn’t surprise me if it made a pleasing but nonstrategic amount of money, because the VC crypto spasm didn’t stop till this year.
Now in November 2022 I can’t imagine that service has a bright future, but I’ve been wrong before.
Coda: Property marks · At some point toward the end of this story, I got a call asking me to come down to Seattle. An important customer was coming in for a few days at the Executive Briefing Center; they wanted to hear about Blockchain and someone pointed them at me.
The customer was a very major international agency that you’ve heard of; not sure if they’re big AWS users or were just kicking the tires. In any case, they were being taken very seriously; presentations from a dozen or more product teams stretched over two or three days.
I huddled with the sales team lead and disclosed that we weren’t planning to ship anything big and that the engineering leadership was skeptical of blockchain in general. I wondered if it would be OK to share that. She looked troubled and said “That’s a judgment call and I can’t make it, you have to. But listen first.”
So I made a presentation about the virtues of ledgers and of cryptographic signatures, mentioned Byzantine generals, and noted that the area was extremely interesting but somewhat unproven in practice.
The customer CIO, an extremely smart person, spoke up, in beautifully-rounded European vowels: “Here’s a use case I’ve been told about that’s on my mind.” He named a region in Asia and explained that the small farmers there mark their landholdings carefully, but then the annual floods sometimes wash the markers away. Then unscrupulous larger landowners use the absence of markers to cut away at the smallholdings of the poorest. “But if the boundary markers were on the blockchain,” he said, “they wouldn’t be able to do that, would they?”
Dear Reader: What do you think?
I thought. Then said “As a lifelong technologist, I’ve always been dubious about technology as a solution to a political problem. It seems a good idea to have a land-registry database but, blockchain or no, I wonder if the large landowners might be able to find another way to fiddle the records and still steal the land? Perhaps this is more about power than boundary markers?”
Later in the ensuing discussion I cautiously offered something like the following, locking eyes on the CIO: “There are many among Amazon’s senior engineers who think blockchain is a solution looking for a problem.” He went entirely expressionless and the discussion moved on.
P.S.: What now? · I still think about those boundary markers.
Up until this year, most of my conversations with crypto promoters have had problems. For example “This person thinks proof-of-waste is OK and thus can be safely ignored.” And “This person is pretending that the proportion of crypto biz that is actually Ponzi doesn’t round to 100%.”
Bitcoin itself can rot in hell and I hope it implodes tomorrow. But the Eth folk managed to get proof-of-stake to work at scale; good on ’em. Still, the explosions of non-Btc crypto Ponzi schemes continue, in a boring and discouraging rhythm.
I’m not prepared to say that no blockchain-based system will ever be useful for anything. But I’m gonna stay negative until I see one actually at work doing something useful, without number-go-up greedheads clamped on its teats.
And I’m glad AWS didn’t make a big bet on it back then.
And I’m sorry for the folk at the Australian Stock Exchange who tried so hard for so long.