The Public Stack Summit

The Public Stack Summit took place on a boat. This was a great idea — not only were we all there the whole day, there wasn’t any wifi and once we were out in the bay there wasn’t really any mobile internet either. So everyone paid attention.

Public Stacks Boat

We started out by talking about what we hoped a public stack might be. That meant discussing both ‘public’ and ‘stack’, but was oddly useful. Waag and Decode had convened a really diverse group, which managed to include an incredible range of tech builders at (almost all?) stack levels, as well as both purists and pragmatists. Because everyone was deeply into this space, we didn’t need to set the scene or even describe the problem. I have far too rarely been in a room with people able to knowledgeably discuss internet energy consumption, RISC V, the challenges of consumer electronics manufacturing, the dominance of Cisco/Juniper/Huawei in switchgear, issues around accessibility, inclusion and design in free software, discovery in decentralised systems, illicit capital flows, open source sustainability, nation state threats to the internet, the need for a new narrative to explain these ideas. We touched on enemies too rarely heard in the sudden excitement about Facebook and Google — what about SAP in the public sector? The issues around how SMTP systems have evolved?

It was really great to explore the deep technology side of things, and not just have a somewhat confused discussion about various things that concern us about Facebook or algorithms right now.

And unusually for an event of this type, I think we actually made progress. We had different ideas, but there was enough of a shared vision to suggest perhaps a movement could emerge. It was possible to discern some next steps; there were definite specific collaboration ideas being discussed between participants; and we recognised (albeit quietly at times) some of the practical realities that this work faces. We managed to avoid getting sidetracked into whether any change was possible without getting rid of capitalism entirely.

Still, rebuilding the internet from the ground up would be a big project. What are the critical places to start prototyping and building? What should we look at first? What places in the stack, or what application areas that cut through the stack, are high leverage opportunities, that could move things towards our shared aims? How do we get more systemic interventions, like Fairphone, which can change things on multiple levels?

What are the key public, common bits of a stack? What are bits where more variety and competition would be OK? Which parts of the stack need intentional design to avoid corporate or monopoly capture, or damage from bad actors?

What are the values we would want a new internet to embody? We had lots of nice ideas. Perhaps more importantly, where are we willing to make compromises, given it seems unlikely that we can have all the things we want in a reasonable timeframe?

We have to remember that not everyone shares our dreams. There will always be people who seek power, and who do not share our ethics. And we are starting from here. A time of internet centralisation not just with Google and Facebook and Amazon in the US, but Tencent and Alibaba in China, and not just at the social layer, but down through the dominant players in compute, in silicon, in the network infrastructure. A time when AI is dominating the global narrative, and dividing the world into competing factions — USA, China, and Europe. (Not to mention other powerful groups who see cyberspace as a place to fight over and within. And what about all the other countries? Presumably they are stuck in an impoverished state, or have to choose one of the big three to follow…) I don’t know what I think about calls to take back democratic control of the internet, in a time where democracy is ending.

What would a technology agenda look like if we started from genuine concerns of large numbers of people today? If we started from general civil society interests, not from an internet perspective? Other (most) people are likely to be thinking about other problems. Should we prioritise them?

Thanks to the amazing team at Waag for inviting me. This was one of the most inspiring and thought-provoking events I’ve been to in ages.

My notes follow, reflecting my particular interests (although not necessarily my views), and written without much editing :)

What is the public stack? Let’s start with: what do we want ‘public’ to mean?

It’s something like our values. It’s more than ‘privacy’.

It’s about Human Rights — real, physical rights, that should be valid in the digital realm, not replaced by weaker online equivalents. We discussed how much agency we have to secure such rights; do some smart city projects assume that their democratic mandate means they can decide what citizen data should be used for, without individual control?

We talked about whether ‘citizen’ was right. “User” is definitely wrong. But not everyone is a citizen. Perhaps GDPR is an inspiration here, with the idea of a “natural citizen”?

We value the commons, collective goods, alongside person-centric design and personal control (not just over our data, but over our software, our hardware, our network infrastructure).

Is self-sovereignty a prerequisite to have a commons?

Sustainability, in energy and materials, came up a lot. Is this a value we want to embed in our new vision of the internet? It may go together with human rights — especially rights for the workers in mines and assembly plants who create our electronic goods today. But these are hard things to change. And hard to relocate into Europe today, perhaps.

Or does public simply mean “not in private hands” ?

The ‘stack’ had perhaps a dozen layers in our final conception of it. These were not at all the same as the GNU stack, which we had on a poster; many of us gazed at it, trying to decipher its colours and meaning. This was too complex to be useful to us.

Bad photo of Possible GNU Internet poster

It would probably be useful to map a few different technologies to this stack framework. Were we missing anything, perhaps around machine learning, compute, cyberphysical systems?

I also wondered if ‘stack’ was too simplistic a concept for a truly networked era. But perhaps we need to stick with simple views of things, if we want to be able to talk about different futures across technology, policy and other communities. Perhaps the purpose of the stack is to enable better conversations — so it’s less a stack as such, and more of a framework to explore different ideas.

Everyone seemed to like Raworth’s ‘doughnut,’ used to reimagine economics; so perhaps we are looking for something similarly iconic.

It was good to see real p2p — in the sense of point to point — networking discussed, not just ‘decentralisation’ in a vague sense.

What we discussed in the final session had layers — technical, governance/management, values. And we added one more — financial, capital, perhaps “euros”, to reflect that resourcing was clearly critical.

I didn’t manage to get a photo of this final stack, with pancake-layers, and a more detailed technical stack layer list. You can see some of it here:
Boat photo showing work in progress public stack framework; photo from Waag —

Reclaiming economics is part of the same movement as the public stack, perhaps. We need to measure the right things, too. We don’t know what the right things are, yet.

We tried not to say “business model” but “resourcing model” to reflect that not all parts of the stack might be resourced commercially. Some commons things may be utilities, funded from taxation or as club goods; it’s not all business. (Or not all good business. We did have industry in the room, too, who pointed to the tension between offering dumb pipes, which are brutal commodity businesses, and ‘adding value’ by removing the neutrality and integrating with services further up the stack.)

How pure is our financial backing? We are not replacing capitalism here, so we still need money for food and rent. What strings are attached to the money taken by nonprofit projects and NGOs? What about the invisible strings? Must we refuse money from Google? Should we shun Mozilla for taking it — even if it unlocks large scale work? Is any corporate money a hostage to fortune?

A public stack perhaps is most useful as a tool to explore what good looks like, across more than just technology at one level.

That was certainly the case in the bar afterwards, where we used this image:

Bad photo of the bar-era Public Stack sketch, showing how it might work for Chinese, European and Silicon Valley tech values

Is the dream of a fully public stack too utopian? Europe faces challenges on trade and industrial strategy and more, and these interests may make our vision simply unrealistic.

Someone suggested that perhaps we would have to compromise, with a sovereign stack, rather than a public one. I initially misinterpreted this as self-sovereign — but I think sovereign was what was meant. A European set of technologies, not colonial tech from the US (or China). Services like AI increasingly drive the whole tech stack, down to the silicon. Maybe we don’t get to have the privacy and control we would like individually, because of the need to compete at least somewhat in sectors like AI, where data is fuel. Maybe we actually need the billions of corporate investment we can only get in a sovereign stack, not a public one.

We acknowledged the people who not only were not on the boat, but for whom ‘making ethical choices’ is too much amidst challenging, busy, pressured lives. We cannot only build for the privileged, educated audience who can afford to pay for ethics, privacy, control. As Europeans, we said: this is where government has a role, helping foster an environment where people don’t have to pay more or figure out high risk areas. (I guess other countries might look to civil society, or charity, or other groups to help with this?) “Managed for everyone, not managed by everyone” — we need to get away from a narrative of ‘self hosting’ and taking personal control.

Many of the concepts we talked about are simply not being discussed in the policy world today. Alternatives such as keeping data local and learning at the edge, or peer to peer resilient systems, are just not on the radar there.

It’s important to remember the voices less heard, whether through creating an inclusive environment to develop technologies, or thinking about accessibility and designing for different communities. Or both. Whatever the public stack might be, it should be developed with inclusion in mind, perhaps specific activities and standards. We touched on how it sometimes wasn’t enough to do this ‘later’ or when a project hits scale; perhaps projects with the intent to be part of the public stack should build inclusion in from the start, at an architectural and vision stage. So “public code” might be more than just open source (or free software), but have inclusion, accessibility, human-centric design integrated through it.

I felt, overall, that we were talking about rebuilding the internet, although some people said we weren’t. There was a sense that we don’t even understand what we have in the internet now — the hidden state and corporate power centres, and what they do, are not clear to us. Who are we to rebuild it anyway? It was people not unlike this group who created the 1990s narrative and vision of the internet, and who failed to foresee or prevent its capture by other interests.

In any case, whether we are re-engineering the whole stack or just parts of it, we will need to find ways to move or “bridge” people, information and systems from today’s internet to tomorrow’s.

What else remains?

Values can be easily missed. In the early days of Flickr, values were archival, creative commons, sharing; a big contrast to today’s ephemeral Snapchat, Instagram. It’s easy to forget values, unless they are very clear and overt.

I was surprised to hear someone from the decentralised, democratic civic space say that industry had to be involved, because otherwise you couldn’t have reliable software. An unexpected perspective. This made me wonder whether some parts of the FOSS community are good at engaging civil/civic folks, but perhaps don’t demonstrate the rigorous engineering of other bits of the FOSS community…

The idea of “control and convenience” as a core concept seemed compelling. Convenience because it’s an essential aspect of showing an alternative to today’s internet systems. Control is perhaps more a value of this community, not something the general public would demand. We wondered whether “community” was an important addition, to capture some sense of collective value, not just personal, self-centred things.

It’s been a while since I’ve heard arguments for free, not just open source software.

It’s too often forgotten that there are people you don’t want in an online space, or in the community designing tomorrow’s internet. Psychopaths, and people who enjoy disrupting and destroying things. We must remember that they exist and think about how we design our communities and spaces in light of it.

We will need to use adversarial design practices, not just for individual applications, but at an architectural level.

There will be an IGF session on funding the free and open internet in November, convening all the funders interested in this space. That seems useful. There will also be funding from the EU next generation internet — 3 pots of funds for different topics each year.

Procurement rules are powerful. What can we specify through them?

Perhaps we should map out the European space — organisations, funders. Remember we do have great stuff too, it need not be eclipsed by the USA. We could strengthen our work by raising the profile of local endeavours, and resources. We could use this to spot gaps, where a homegrown solution might be possible (or where we may benefit from a local equivalent or partner to a US project). We could map the bad things as well as the good; recognising that a complete picture may be better than agonising over what criteria we use to rate projects. (Even Mozilla may not be 100% good, if we are doubtful about the Google funding!)

What about the global South? We may be neglecting them by focussing on Europe. Or, is the challenge so hard, that we must focus on Europe alone and what can be achieved here?

Can we disrupt routing centrality, by making it easier to route between ‘nice’ ASNs (those of universities, public sector broadcasters, etc)? Could this be an economic advantage for developing countries, whose traffic goes through countries they may not want to align with?

We don’t actually have a good map of how the internet looks. it would be useful to flesh out the state of the internet through the stack now, maybe for here and for China and the USA. And to track changes over time…

No one wanted to advocate blockchain as a solution, for anything.

Who is campaigning to stop bitcoin, on environmental grounds?

Is no one campaigning and working on hardware issues from a systems perspective, looking at patents, manufacturing, firmware, security etc? There seems to be lots of niche work, but nothing bringing it all together. An opportunity for some useful convening?

Why are our open source developers so surprised and frustrated that Github was bought by Microsoft? The VC funding model meant some sort of commercial exit was almost inevitable. Are Microsoft seen as bad guys now, or is this cultural memory from the 1990s? Are such trivial concerns, built on shallow conceptions of the world, the only foundations we have to build a better future upon? Or are they just the surface fluff that comes from sampling the will of communities like open source — where we’d hope to find allies — in a social media, ‘hot take’ era?

The most succinct summary of the problem: It’s everything from warlords controlling the mining of key minerals we need for computers and phones, to Airbnb driving rents in the city beyond what local workers can afford.

What would a privacy-preserving and participatory smart city look like?


Engineering useful new technologies & systems. Gets stuff done. Doteveryone, Past: Cambridge Makespace, Open Knowledge, Evi, AlertMe, AT&T Labs, & U Cambridge.