We're unveiling Aster, a suite of Atmosphere1 services designed for researchers and academics, at IOSP 2026 on October 12. We're in alpha mode, providing an invite-only service, because we have some difficult, highly interconnected questions to answer. And we'd appreciate your help.
In a nutshell, Aster will introduce new trust signals into online conversations by providing researchers and academics an Aster identity and other Atmosphere services. As our opening post put it, better trust signals online means better science:
"The last few years have not been great for science. Public trust is at historic lows, and research is straining under mounting pressures — not least the way we publish and communicate science, which is long overdue for an upgrade" - Announcing Aster - a home for researchers on the open web
Questions about verification
That mission immediately raises the following questions:
Who counts as a "researcher or and academic"?
How do they prove it?
Who decides, and How?
Is there an appeal process?
What exactly are we identifying - for example, what does it mean if an Atmosphere account has an "name.aster.id" handle?
that they're a human?
that they are who they claim to be?
what about research organisations? Or bots run by scientists or universities?
How do we indicate any of this online? A trust signal isn’t much use if nobody sees it.
These are all questions about verification. This has been a hot topic since 2022, when X introduced pay-to-verify, and hotted up further this year with WSocial's "upload your ID card to our (leaky) server" and the EU’s recently proposed KIDS Act, which imposes age assurance tools on everyone opening a new account.
In science, verification has additional issues: science is full of gatekeepers who determine who is and who isn't a scientist, who gets and who doesn't get funding, and what gets funded. Do we want to reinforce those structures? Maybe not, but how can we feasibly offer something else?
Aster's goal is to introduce more trust signals online, so these questions must be answered clearly (easy to understand) and transparently (nothing to hide). Moreover, those answers must be defensible (well thought through) and credible (we have the resources to implement).
Which is why we want your thoughts, and explore these questions a little further, below.
Starting positions
Let’s start with what we came up with for launch day.
At a glance:
We want to be open to self-identified researchers, but we also see Aster providing multi-level trust signals
While we don't want bots, we also don't want a proof of your humanity via a scan of your ID card
Aster therefore needs to accumulate trust signals from a variety of sources, and will explore empowering organisations to provide them
We intend running a labeller to display those trust signals, and will explore verification badges, although that might require our own app
Aster will need reporting and appeals processes, and those cost money we haven’t yet raised.
Until we've developed not just the answers but also the processes and tools those answers imply, we'll keep Aster invite-only. That means:
What an Aster handle indicates
Right now, an Aster identity means that when the account holder asked for it they confirmed that they are a researcher, and we saw nothing to doubt them when we made a first check.
If your reaction is "that's not much", you're right. But that's the point: while a manual check is not a very strong trust signal, this is only the beginning. With further development, Aster will be able to display additional trust levels, so it's normal that earning today's base trust level is not impossible to earn.
Fundamentally, we need to choose a preferred failure mode: do we prefer to accidentally let in someone who shouldn't be on Aster, or mistakenly keep out someone who should? We're currently going with the first: science already has an abundance of gatekeepers, and most of it falls on the people with the least institutional cover. So we're starting permissive, and will build the trust layers on top, as set out below.
Having said that, if manual checks are the mouth of Aster's funnel - the step you must pass through before other trust signals can be applied - this clearly won't scale without "invite chains", which have their own issues (next).
Layered, not gated
Interesting things become possible once you have an Aster account and PDS, as we're exploring running a labeller that can mark accounts with higher levels of trust: this person has been vouched for, this person is known to an organisation we trust, who checked their ORCID number, etc...
In particular, universities, research organisations and funding agencies already know their researchers: rather than Aster trying to assess tens of thousands of strangers, these organisations could confirm the people they employ or fund.
This is why Aster's trust signals are layered, rather than gated: an unaffiliated researcher won't be locked out of Aster because no institution will vouch for them, but those verified by their institution will be labelled as such. Moreover, "invite chains" provide a traceable chain of responsibility: if someone causes harm, we can see who vouched for them.
Having said that, there are a few things we should keep in mind:
invite networks inherit the exclusions already present in academia ("you'll get in if you know someone already inside"), so they're not the only solution we should develop
labellers are only seen by people who subscribe to the label, so maybe we'll need our own app?
We're not asking for your ID
We've ruled out document-based identity verification for now, as it creates a privacy liability we are not resourced to handle. It is also exclusionary, while we want to be open.
Other projects are working on interesting verification technologies using "zero knowledge proofs". where you can provide proof of an attribute (age, qualification, etc.) without releasing any other information, let alone uploading ID scans. One of the Aster team is involved in one of these projects with Eurosky Social, so we'll take a look at this when it matures.
Share your thoughts, get involved
We have a lot to build: a labeller, vouching processes, systems for reporting problematic accounts, an appeals process, and more.
But most of those tools and processes will be built to support our answers to the above questions. We need to be sure of those answers first. And for that we need your opinion, so:
comment on this post using the form below, or use Leaflet's clever "highlight, share and comment" feature to post your thoughts on a particular section via Bluesky
reach out to us (see FAQ: How do I learn more and contact you?)
- Mathew