Trust and trustless: a tale of one of the first adversarial interactions within Arweave
Hi there guys, some peculiar events made me return earlier from my hibernation period, so here you have it, my first article in ages.
If you are asking what peculiar events, then I know for sure that this article - for one reason or another - had a bigger reach than usual because otherwise, only an individual familiar with the Arweave ecosystem will access this link, and those guys definitely know what I'm talking about. (Trust me bro, I know our audience).
However, let me swiftly present what happened to whoever doesn't know what I'm talking about: Sam, Arweave's founder, posted this tweet:
Irys responded shortly with this:
Now, both Sam and Irys come with some follow-up tweets, you can find them if you access Twitter, X, whatever. The idea is that Sam's initial message and the response from Irys generated a weave of reactions throughout all the social platforms.
I didn’t want to publish back then; rarely you can have clarity in the heat of the moment. However, I tried to talk with as many of the involved parties as possible. Notably, I managed to have a call with Sam and the founder of ar.io, @vilenarios, in the very early hours after Sam's announcement. Josh, the founder of Irys, told me he'd probably be able to chat later this week, so I don't really have his reaction from those heated moments.
Now, everything appears to have settled. Hopefully, instead of rage-driven reactions, common sense reigns again. Still, a lot of things were said some days ago, and I think that it makes sense that at least only a couple of them should be addressed as calmly as possible.
What was the process that led to taking action against Irys?
As you can see, I do not even mention the "reason" that led to the origin of this process because I would argue that the process itself may be even more relevant in the context (now, keep in mind that I'm not a native English speaker and I'm trying to use the meaning of the words as clear as possible, because the nuances are important).
So, initially, the potential "reason" came to the knowledge of Forward Research. They started to talk firstly with Irys. If the dialogue would have reached a consensus, the process would have ended there. Clearly, a consensus was never reached back then, so Forward Research started to talk with relevant teams from the ecosystem, and this second talk actually reached a consensus. In this second talk, everybody considered the "reason" sound enough to move forward and take action. Obviously, if the second talk hadn't reached a certain consensus between multiple teams from the ecosystem, Sam's message wouldn't have existed in the form it reached the public. The second talk could have had multiple resolutions: the teams contacted by Forward Research could have reached the conclusion that the "reason" is not sound enough to take action, or even to side with Irys. However, those potential endings didn't materialise.
Contrary to what some may tend to believe, the process couldn't have happened like this: Forward Research and, subsequently Sam, found out about the "reason" > Sam unilaterally took action immediately. Why? Because the envisioned action wasn't Sam's to take from a technical and practical standpoint (we will get there).
Was an initial action taken against Irys, by whom?
As condensed as possible: NO; no action was actually taken against Irys, besides the tweet itself. However, there was a warning, that Irys's access to the optimistic cache, and data indexing to the arweave.net gateway could be stopped.
If this privileged access is cut off, what does this truly mean? That, if you are currently using arweave.net to retrieve information from the weave, you won't be able to see the files bundled with Irys in the first 20 minutes from the initial upload to the network.
To summarize: the action would have consisted of cutting the access to a single feature linked to a single gateway. To have a scale reference, take a look at this view available on Viewblock; there are over 200 gateways active at the moment of writing this article.
Also, what really is this optimistic cache access and how does it function right now? Well, you know that I'm not a very technical person, so I'd use a rather simplified explanation, hopefully, a correct one: the process of bundling data from multiple sources, committing it to the network, and being sure that the data settled onto the chain and is seeded to the miners, takes time, call it a lag - 20 minutes or so.
As a user, you may want to have instant access to what you uploaded to a bundler. What the optimistic bridge does is that it lets a bundler service communicate directly to a gateway and transform the bundler in some sort of mempool - basically, for the first 20 minutes, the gateway will show the data received directly from the bundling service, having some sort of certainty that the said file will be uploaded to Arweave.
So, in order for this feature to function, it needs to have two entities: a bundling service and a gateway, and between them should exist a certain type of relationship. If this relationship doesn't exist, the feature of optimistic uploads can not happen. The relationship can be based on trust or can be trustless. In this specific case, the relation between the said gateway (arweave.net) and the bundling services that were granted access to upload data optimistically was based on trust. The trust of the gateway that the bundling service will certainly upload to the Arweave network 100% of the data that was fed to the gateway. (I will also explain why, in my vision, at the moment, it functions on trust)
Now, as you may have already understood, access to this feature, in its current form, is not an unalienable right derived from a trustless interaction but a privilege gained and maintained by trust. When the trust is broken, whatever the "reason", the outcome is pretty predictable.
Ok, let's say the "what" was some sort explained, but by whom would have been this action actually taken? By Forward Research, by Sam himself? No, Forward Research couldn't do it by themselves; they needed the consent and the action of the ar.io team that has been stewarding arweave.net for the entire @arweaveeco. Even if Forward Research is the owner of arweave.net, the entity that operates and pays for the optimistic bridge of arweave.net is ar.io.
Why is this important? Because it shows that the action is taken via consensus of multiple parties; it isn't the result of unilateral will.
The paradox
If the action against Irys will ever happen, it’s only because of Irys - and I’m not talking here about what they’ve done and they shouldn’t, but about what they didn’t do, and they should have done.
Let me explain: initially, Irys was called bundlr.network and from my understanding that ".network" had a lot of meaning. When the ecosystem developed first the ANS-102 (very short-lived) and the ANS-104 (a version of ANS-102 in binary iirc) standards, everybody expected that bundlr.network would not only be a bundling service, but would evolve to become a trustless network of many bundlers.
bundlr was expected to do for bundling data and serving it through optimistic bridges to gateways, what ar.io managed to do for the gateways themselves: to create a network/protocol, that would decentralise this aspect. If I got it right, that network could have been working by letting any bundling service stake a number of tokens in order to have access to the optimistic bridge of one/many gateways. That bag of tokens would have been slashed if the said bundling service hadn't respected the network rules.
If this piece of trustless tech had been created, nobody, not even the combined will of the ecosystem, would have been able to remove Irys' access from the optimistic bridge, as long as Irys had been respecting the rules of the network.
There is a tragic irony at play indeed: probably bundlr/Irys, was the project in which the ecosystem invested the most amount of trust during time. Trust that they'll fulfil this mission towards a trustless optimistic bridge. While time passed, and the trustless infrastructure was still late to appear, trust eroded bit by bit. This led to frustration - you can clearly feel it in this tweet, for example:
Epilogue
We don't have tyrants within Arweave.
The crushing response that came from almost all the projects from the ecosystem wasn't a sign of showing "allegiance" to Sam. Sam's "power" is directly proportional to his alignment with the general sentiment of the ecosystem. My take is that the trust towards Irys started to erode within each project, way before this event. The "reason" was just a fuse that ignited a powder keg that was slowly filled over time.
Before even trying to talk about exploring forks, or evolutions of the network, before trying to get what is an anti or pro-social way of doing it, I think that maybe Irys should try and create that open-sourced and trustless piece of tech that was awaited by many. It could be the first step on the path of regaining trust.
