Andras Gerlits
@omniledger.io
I have built the first async, consistent data-platform
https://omniledger.io/
I build distributed systems @Citi
I also write about distributed systems
https://medium.com/@andrasgerlits
https://omniledger.io/
I build distributed systems @Citi
I also write about distributed systems
https://medium.com/@andrasgerlits
There's a great interview with Leslie Lamport, in which he makes the point that any time-ordered system will contradict the observed (causal) order of events as seen by different processes in different places. He's clearly right about this, except for his conclusion.
November 16, 2025 at 6:00 AM
There's a great interview with Leslie Lamport, in which he makes the point that any time-ordered system will contradict the observed (causal) order of events as seen by different processes in different places. He's clearly right about this, except for his conclusion.
Time is experiencing changes. In distributed systems, we focus on composing these changes eagerly into a unified view for any potential observer. This eagerness is the bottleneck. If you just lazily calculate each observer's current view, the known problems just don't apply.
November 15, 2025 at 5:46 AM
Time is experiencing changes. In distributed systems, we focus on composing these changes eagerly into a unified view for any potential observer. This eagerness is the bottleneck. If you just lazily calculate each observer's current view, the known problems just don't apply.
The basic facts of physics tells us why distributing systems is hard. It's because different time-coordination systems move at different speeds. The only logical conclusion here is that these need to be calibrated to each other if we want to establish order between them.
November 2, 2025 at 4:34 AM
The basic facts of physics tells us why distributing systems is hard. It's because different time-coordination systems move at different speeds. The only logical conclusion here is that these need to be calibrated to each other if we want to establish order between them.
Second outage in two weeks. I guess it's time to be educated about how nobody could do better than be down for half a day again.
October 30, 2025 at 9:12 AM
Second outage in two weeks. I guess it's time to be educated about how nobody could do better than be down for half a day again.
The AWS outage has nominally happened because of DNS, but the root cause is that we centralise data-collection and management. If data was delivered to observers from multiple sources (with a shared timeline), central outages cannot happen. Not all reactors are Chernobyl
October 26, 2025 at 7:05 AM
The AWS outage has nominally happened because of DNS, but the root cause is that we centralise data-collection and management. If data was delivered to observers from multiple sources (with a shared timeline), central outages cannot happen. Not all reactors are Chernobyl
The consensus in IT that cloud providers failing centrally is still the best of all possible worlds is not only self-serving but also false. It leads to clients not asking questions so everything staying the same. I think the reason developers excuse this is because 1/x
October 24, 2025 at 4:15 PM
The consensus in IT that cloud providers failing centrally is still the best of all possible worlds is not only self-serving but also false. It leads to clients not asking questions so everything staying the same. I think the reason developers excuse this is because 1/x
People offload to cloud-providers, not because that is the technically superior thing to do, but for the same reason they hire McKinsey, to shift blame from themselves.
October 21, 2025 at 6:47 AM
People offload to cloud-providers, not because that is the technically superior thing to do, but for the same reason they hire McKinsey, to shift blame from themselves.
I love how absolute science is. It's the only area where you can be unapologetically certain about being right even if everyone else is telling you that you're wrong.
October 21, 2025 at 4:36 AM
I love how absolute science is. It's the only area where you can be unapologetically certain about being right even if everyone else is telling you that you're wrong.
Wrt today's failure: There were two things that surprised me very much after I started working in software innovation. The first was how receptive academia was to new ideas. I'm very proud of the fact that so many took the time to understand how we solved the fundamental 1/x
October 20, 2025 at 6:37 PM
Wrt today's failure: There were two things that surprised me very much after I started working in software innovation. The first was how receptive academia was to new ideas. I'm very proud of the fact that so many took the time to understand how we solved the fundamental 1/x
Some AI exec: I've seen a future version of our product and it's so powerful, it scares me.... (stares off in the dark with a tear in their eye). I wake at night, trembling for the future of humanity...
Reporter: Can we see it?
E: No, it's too powerful!
Reporter: Can we see it?
E: No, it's too powerful!
October 14, 2025 at 12:06 PM
Some AI exec: I've seen a future version of our product and it's so powerful, it scares me.... (stares off in the dark with a tear in their eye). I wake at night, trembling for the future of humanity...
Reporter: Can we see it?
E: No, it's too powerful!
Reporter: Can we see it?
E: No, it's too powerful!
Call me a Luddite, but I don't think it's great that the tech-sector is beholden to a small, entrenched group of VCs. It's ironic that "incumbents only do innovation theater" used to be their rallying cry.
October 14, 2025 at 5:48 AM
Call me a Luddite, but I don't think it's great that the tech-sector is beholden to a small, entrenched group of VCs. It's ironic that "incumbents only do innovation theater" used to be their rallying cry.
'm convinced that the love for CAP, Kubernetes and microservices come from the same place. CAP was taken to mean that you can't ever have a perfect platform, so might as well cook your own.
Simplicity is the last thing they want. They would much rather feel clever.
Simplicity is the last thing they want. They would much rather feel clever.
October 13, 2025 at 7:10 AM
'm convinced that the love for CAP, Kubernetes and microservices come from the same place. CAP was taken to mean that you can't ever have a perfect platform, so might as well cook your own.
Simplicity is the last thing they want. They would much rather feel clever.
Simplicity is the last thing they want. They would much rather feel clever.
A fundamental principle of spacetime is that the further away you are from a source of information, the longer it will take for it to travel to you. This increased time means your local time can progress faster, so the remote clock will have "bigger chunks of data".
October 11, 2025 at 3:43 AM
A fundamental principle of spacetime is that the further away you are from a source of information, the longer it will take for it to travel to you. This increased time means your local time can progress faster, so the remote clock will have "bigger chunks of data".
Krasznahorkai's work _is_ the malaise carried by Hungarian intellectualism. In his interviews, he makes his loathing for Hungarian nationalism very clear, but I don't know a single serious thinker here (in Budapest) who doesn't share his alienation.
When Imre Kertesz won the Nobel in literature in '02, almost nobody knew his work. As for Krasznahorkai, it's a different picture. Almost everyone I know read something from him at some point and certainly know his work with Bela Tarr.
Literature is only half-dead.
Literature is only half-dead.
October 10, 2025 at 4:05 AM
Krasznahorkai's work _is_ the malaise carried by Hungarian intellectualism. In his interviews, he makes his loathing for Hungarian nationalism very clear, but I don't know a single serious thinker here (in Budapest) who doesn't share his alienation.
When Imre Kertesz won the Nobel in literature in '02, almost nobody knew his work. As for Krasznahorkai, it's a different picture. Almost everyone I know read something from him at some point and certainly know his work with Bela Tarr.
Literature is only half-dead.
Literature is only half-dead.
October 9, 2025 at 3:07 PM
When Imre Kertesz won the Nobel in literature in '02, almost nobody knew his work. As for Krasznahorkai, it's a different picture. Almost everyone I know read something from him at some point and certainly know his work with Bela Tarr.
Literature is only half-dead.
Literature is only half-dead.
I don't see many posts of people "buying" innovation, yet people talk about "selling" it all the time. Why? I don't think people give nearly as much thought about buying as sellers, who are drawing out careful plans on selling them.
October 9, 2025 at 7:08 AM
I don't see many posts of people "buying" innovation, yet people talk about "selling" it all the time. Why? I don't think people give nearly as much thought about buying as sellers, who are drawing out careful plans on selling them.
The confusion comes from the model. A process is data processing data, so it's not (necessarily) a source of information. The jury is still out, but I think people are (original sources). Generals can "think of things", processes can't.
Two generals posits that we have nodes which can't reliably communicate, because of the noisy channel between them, where packets can get lost and the node can't tell a lost from a successful packet. This posits a few things. First, that there's exactly one instance of each node.
October 9, 2025 at 2:48 AM
The confusion comes from the model. A process is data processing data, so it's not (necessarily) a source of information. The jury is still out, but I think people are (original sources). Generals can "think of things", processes can't.
Absolute time is a bad idea because nodes on our grid can become isolated, in which case other nodes can't know if they are missing data, ie: "are consistent".
A relative clock, disallowing isolated nodes from progressing the global timeline work around this.
This isn't magic
A relative clock, disallowing isolated nodes from progressing the global timeline work around this.
This isn't magic
October 8, 2025 at 2:53 PM
Absolute time is a bad idea because nodes on our grid can become isolated, in which case other nodes can't know if they are missing data, ie: "are consistent".
A relative clock, disallowing isolated nodes from progressing the global timeline work around this.
This isn't magic
A relative clock, disallowing isolated nodes from progressing the global timeline work around this.
This isn't magic
With hindsight, I'm surprised the Wright brothers didn't need to build an entire commercial airline before they were taken seriously.
October 8, 2025 at 9:15 AM
With hindsight, I'm surprised the Wright brothers didn't need to build an entire commercial airline before they were taken seriously.
A shared key between two distant locations can't be updated faster than the one way communication between the two, say 120ms
You can trade fairness of access for update speed. Placing a key somewhere means prioritising local updates over remote ones.
You can trade fairness of access for update speed. Placing a key somewhere means prioritising local updates over remote ones.
October 7, 2025 at 2:46 AM
A shared key between two distant locations can't be updated faster than the one way communication between the two, say 120ms
You can trade fairness of access for update speed. Placing a key somewhere means prioritising local updates over remote ones.
You can trade fairness of access for update speed. Placing a key somewhere means prioritising local updates over remote ones.
The problem with the "Two Generals" is the same as CAP or PACELC: they start from overly restrictive conditions. We know since Paxos that we can have multiple copies of the "same" general and our paper explains how to freeze partial/subjective time for decentralised observers
October 1, 2025 at 7:02 AM
The problem with the "Two Generals" is the same as CAP or PACELC: they start from overly restrictive conditions. We know since Paxos that we can have multiple copies of the "same" general and our paper explains how to freeze partial/subjective time for decentralised observers
I wrote a short article about why clearing is going to be a fundamental problem for stablecoins and why established academic thinking tells us that this problem is unsolvable. The next article will explain the solution.
andrasgerlits.medium.com/why-clearing...
andrasgerlits.medium.com/why-clearing...
Why Clearing is a Distributed System Problem and Why That’s Bad News for Stablecoins
That all financial transactions rely on trust is nothing new. Just how far we need to trust the other party however, makes all the…
andrasgerlits.medium.com
October 1, 2025 at 5:02 AM
I wrote a short article about why clearing is going to be a fundamental problem for stablecoins and why established academic thinking tells us that this problem is unsolvable. The next article will explain the solution.
andrasgerlits.medium.com/why-clearing...
andrasgerlits.medium.com/why-clearing...
When coding, it feels like we're creating processes that evaluate their data and makes decisions, but that's just the way we update shared data. Seamless data-sharing is therefore the best possible way to do microservices, orchestration or any kind of integration
September 30, 2025 at 10:57 AM
When coding, it feels like we're creating processes that evaluate their data and makes decisions, but that's just the way we update shared data. Seamless data-sharing is therefore the best possible way to do microservices, orchestration or any kind of integration
"Statistical observation is an arbitrary decision on top of an already arbitrary model, so is therefore far removed from human perception."
I think...
I think...
New blog post that's nominally about statistics, based on various things I've been reading lately: surfingcomplexity.blog/2025/09/28/a...
A statistic is as a statistic does
(With apologies to the screenwriters of Forrest Gump) I’m going to use this post to pull together some related threads from different sources I’ve been reading lately. Rationalization a…
surfingcomplexity.blog
September 29, 2025 at 3:05 PM
"Statistical observation is an arbitrary decision on top of an already arbitrary model, so is therefore far removed from human perception."
I think...
I think...
Why lack of global time-coordination is the root (cause) of all evil in our industry
Make Behaviour State Again!
Who creates data? Users, by interacting with systems, which store these on computer-hosts and networks. Are two systems the same if they…
medium.com
August 29, 2025 at 8:00 AM
Why lack of global time-coordination is the root (cause) of all evil in our industry