Audrey Tang is Taiwan’s digital minister in charge of social innovation. Audrey Tang is a software programmer known for revitalizing the computer languages Perl and Haskell and for building the online spreadsheet system EtherCalc in collaboration with Dan Bricklin. She served on Taiwan’s National Development Council’s Open Data committee and…
What can Asian countries learn from the US approach to technology and vice versa?
Audrey paraphrases the President of Taiwan: we used to see democracy as a showdown between two opposing values, but now democracy must be reimagined to become a conversation between many diverse values.
Taiwan is very young, having had its first presidential election in 1996. Because of this, democracy and the internet are intertwined. The Taiwanese talk openly about the “bitrate of democracy” (ex: voting is 3 “bits” of information per person every 4 years). They also talk openly about constitutional design, and just like an operating system they are constantly tweaking it.
Almost every other democracy in the world predates the internet and runs on paper: many practices date back to ancient times when states were run on even lower-tech. The information technology and the tools in use completely shape a state’s approach to governing.
Taiwan’s internet-native democracy is treated as a social technology itself, not just the industrial application of science. If there’s an “Asian” view of democracy maybe that’s it.
What does it mean to be a “conservative anarchist” while working for the Taiwanese government?
Audrey breaks down what they mean when they refer to themselves as a “conservative anarchist”
“Anarchist” is used in its pure meaning: they don’t take or give orders. All the people who work with them are doing it voluntarily, and they with them
“Conservative” is used in a simple way too: mainly, being respectful of tradition. Taiwan has 20+ languages, many indigenous, and these many cultures must be respected and listened to if the whole is to thrive. Instead of making “progress” for one culture at the expense of others, rough consensus among groups towards shared goals.
Another way to put it: Audrey is a true Taoist, but while the western connotations of Taoism may be ritualistic or spiritual, it’s really quite practical.
“I don’t work for the Taiwanese government, I work with the Taiwanese government. I don’t work for the people, I work with the people.” (11:45)
What idea(s) have you found particular useful to your efforts in Taiwan?
All meetings with Audrey undergo a 10 day co-editing period and then are released publicly. This serves to make conversation participants mindful of what they say because they know people in future generations will be watching.
Private meetings allow for people to get away with short-term thinking and proposing policy that is not focused on the long-term thriving of all.
“I think this intergenerational solidarity is built on top of the idea of radical transparency and us contributing to the commons. It’s not just the what of policy-making, but the why and how of policy-making.” (13:08)
The open release of policy-making meetings also serves to get young people, immigrants, and every stakeholder that is not traditionally involved in the processes of government.
Q: does timed release have its place?
Yes, there are some privacy concerns, and anonymization can be done on the recordings during the 10-day editing period. Each participant is responsible for their speech in the video.
Allison: Eric Drexler once said that 20-year or similar timed-release clauses would help to bring some accountability to the government, and could be negotiated down.
Audrey: The reason it’s 10 days and not 10 years is that the private sector is moving at the speed of the internet. If the public sector is moving on much longer innovation cycles it will render itself irrelevant.
As surveillance tech continues to evolve, how can we mitigate abuse?
First off: it WILL be abused. Second: we can just make sure that it’s not the norm through popular demand for alternative solutions. But! Those alternatives must exist to be demanded by the people.
case study: a hostess in the nightlife district got COVID and didn’t report the exposure to contact tracers. Any other East Asian state would probably have forced surveillance or jailed her. Instead, they talked with her.
Turns out, she didn’t want to sacrifice her and the customers’ privacy, a concern the government shared. Instead of forcing it, the government worked to come up with a privacy-preserving solution.
Audrey and the government worked with the nightlife district to develop a system of “participatory self-surveillance” instead of authoritarian state surveillance
Establishments maintained their own private records of who visited, using throwaway email addresses and paper based self-submission. As long as the folks could be contacted, they didn’t need to reveal their names. And after four weeks of no COVID issues at that location, the records are progressively destroyed.
This way the establishments were not driven underground (see the effects of Prohibition in the United States). Rather, we decentralized the surveillance so they can be part of the response team.
“We trust the people more, and trust them to trust each other” (18:52)
Without the alternative this is not possible and a worse solution would be needed
How has COVID affected government technology worldwide?
First off: it’s justified a lot of state surveillance and encroachment of privacy. It’s created unhealthy norms for social sector.
However, Taiwan didn’t need to lockdown at all. No new data collection touchpoints were created in Taiwan and an “infodemic” was avoided.
The Taiwanese people remember the SARS outbreaks and the martial law that was put in place then. They don’t want to do that again.
Taiwan is definitely an outlier in this way though: most other Asian jurisdictions have seen major government overreach.
China claims their incoming digital currency is privacy-preserving, do you know about this?
Most likely the system is pseudonymous just like BTC or ETH, meaning you use addresses to make transactions and don’t necessarily attach your real name to your account.
BUT in the PRC, using the internet often requires real-name ID and facial recognition, so it’s very likely that the government has full identification around the currencies but companies/customers can’t identify others from transactions.
How can other countries engage with democracy as a technology?
“I think what’s most useful is to trust the citizens and apologize swiftly if they find out a mistake has been done and show some real competence by inviting them to work on better alternatives.” (24:34)
case study: The Taiwanese government got the idea to publish real-time data on mask availability at pharmacies. Initially these were being published in documents online, but nobody has time for that.
Some civic hackers built an app to show a real-time mask numbers by tracking the swiping of the Taiwanese National Health cards that were used to receive your mask rations.
But once put in practice, the pharmacies independently invented a “take a number” system and started processing card swipes over lunch instead of real-time, so the map was not accurate, to the point where they even put up signs that said “Don’t trust the app.”
A more authoritarian state may have punished one side or the other.
Instead, Audrey went there and asked: “What would you do if you were the digital minister?” (their favorite question)
They worked with the people running these pharmacies to discover a way to sort of hack the system to remove pharmacies from the map when they were actually out of masks.
The accurate mask map data enabled a clear view of how well the distribution of masks was working, and led the government to adapt its distribution system later in response to the data.
How do you decide to tradeoff between radical transparency and individual privacy? When you are using public data about citizens, when do you expose that?
“We make a very strong delineation between open realtime data and private data” 29:24
Open data like air quality and environmental data is part of the common. This sort of data is commonly collected by kids and citizens as part of educational and public programs to instill a sense of “data stewardship” in the people.
Through becoming familiar with data collection and how it works, data used outside of its main purpose becomes apparent, so the citizens become more aware of and have the vocabulary to push back against problematic data collection.
“It’s like teaching the responsible use of fire at a tender age of six.”
Taiwan is working now on getting GDPR compliant and are also working to increase the use of federated and split learning and homomorphic encryption to learn from their country’s data without revealing identifying information. Because of data competency classes people realize these things are possible and so they demand them.
Like Bucky Fuller said: build a new system to render the old obsolete. Instead of trying to fight against privacy-destroying technologies, they are focusing on building out privacy-preserving alternatives that work for everyone.
Do you have thoughts on liquid democracy?
“Delegating to specific people requires you to maintain constant compact with said people and that makes scaling challenging.” (33:43)
The experiments Taiwan has tried with liquid democracies have been voting for topics instead of people, which has lower cognitive load for people. You subscribe to a set of petitions and get updates so you have input on what matters to you. Faster feedback loops are what we’re looking for and
You subscribe to a set of petitions and get updates so you have input on what matters to you. Faster feedback loops are what we’re looking for and the liquid democracy delegates can actually be bottlenecks on the bitrate of democracy.
Taiwan uses more quadratic voting and quadratic budgeting to move funds to projects, and this way many to many systems where everyone feels like they won because some of their policy goals are met.
Taiwan uses quadratic voting for its presidential hackathon. 5 winners each year get a presidential promise that their policy ideas will become national policy as a reward.
They use the UN SDGs to determine the impact of proposals, where each project must correspond to one or more SDGs. 10 million of 23 million total population vote on the platform for proposals they want, and each person has 99 votes that are dispersed quadratically. The average citizen votes for 5-8 different projects, learning more about SDGs and projects as they go.
Because most of the voters are voting or a collection of proposals that has some synergy, it’s easy to see which teams that didn’t make the cut could join winning teams with similar ideas to keep helping their country, so everyone feels like they won.
Quadratic funding is somewhere in the middle of pure grant-making and pure crowdfunding. Radical Exchange is working with social investment committees in an upcoming social innovation summit to to design that for impact-based investing.
Do you have thoughts on DAOs? other governance technology experiments?
“To me democracy is a form of technology, so as a technologist I’m equally excited about each and every new governance proposal and mechanism.”
When the mature experiments show promise, Taiwan will try them out :)
Taiwan’s Wikipedia page mentions a government sharing economy platform? What’s that about?
No, that’s wrong actually. We developed public platforms for public deliberation ABOUT whether and how to incorporate the sharing economy into the country, not to power it.
What this could be referring to: polis.gov.tw, a “prosocial social media” platform (and a piece of “thoroughly free software” (41:42)) that was developed to help folks find agreement around the sharing economy through the idealogical differences they may hold.
Using the platform, folks voted on agreements around feelings and statements, and avatars move around to show affinity between users. What this reveals is that “most people agree with most of each other on most of the things most of the time” the true shape of democracy:
But on other social media platforms, the divisive statements are the posts that get highlighted: “Private antisocial social media are like night clubs, addictive drinks and private bouncers and all that. There’s room for a nightlife district, but we do public deliberation on public squares.” (42:55)
What technologies are you looking at now?
The idea of democracy as a technology is a key principle. The fact that you can increase the “bitrate of democracy” by using innovation cycles to develop it without needing to destroy the old systems.
What’s stopping more people from realizing this: the false idea that career public servants are resistant to change. This isn’t true: they ARE open to change if it will save time and reduce risks, which these kinds of tools can.
As global institutions continue to lose trust, which emerging attractors for noncoercive coordination do you find most promising?
The idea of data coalitions is getting more attention. People may not call it a coop or credit union, but people are starting to think more in terms of governance and voluntary cohorts
Allison: what are some data coalitions in Taiwan?
~6 million people have downloaded an app that lets you share your national health data privately to integrate with various SDKs that can process and inform people about health matters.
(When) will we have an automated future? What to do to keep humans in the loop?
“We already have an automated present.” (50:32)
“The trick is to automate away things that we don’t want to do and to focus on things that require people-to-people connections because we collaborate much better if we don’t spend time on the chores to enable the collaboration to happen in the first place.” (51:43)
The vote counting can probably be automated, the mechanical parts
“The deliberation to get this agenda on the table in the first place, to find the facilitative space where people can converge on the common ideas and values despite their initially different positions, that’s an art, and what’s an art probably requires humans in the loop and could not be that easily automated, or at least it could be automated to much detriment so we don’t do that.” (51:45)
Allison: What do you think of recent GPT-3 progress?
It’s an inspiring art form, right? It’s this complete sentence art form, and as a poetician I think it inspires poetry. But if you expect a poet, to be your, I don’t know, local clinic doctor or something, that’s asking too much from poets.”
What do you mean by “bit rate of Democracy”?
Anyone can calculate the bitrate of democracy: think of public choices and the frequency at which they are made.
Example: the sli.do platform (where we were submitting our questions for Audrey to answer) allows for a higher bitrate than submitting questions beforehand because you can upvote and downvote and it’s actively shaping the agenda of the meeting without interrupting.
Q: So sort of a measure of agency?
The Polis platform mentioned above allows for 140 character posts that are then resonated with to certain degrees by respondents: this makes for a much higher bitrate than a predefined poll for example, where each participant is only making a series of selections from predefined choices.
How will technology change the global offense-defense balance in the next 10 years?
“I really don’t know about that, because I only work with the people who voluntarily associate with me and they know that I’m working in the radical transparency way. So while my office does have dispatches from around 12 ministries related to, I don’t know culture, public communication, public diplomacy, education, you name it, we do not have any defense ministry people in my office… maybe they’re not that cool with radical transparency. So I know nothing about military, and I don’t really have any context to answer this, beyond what everybody reads in the news.” (55:02)
What is the number one global development or technology you’re excited or worried about?
The idea of data coalitions and data collaboratives. International endeavors that have already helped tackle the pandemic and are now looking at climate change, the infodemic and other big problems
Worried: The tendency to attribute mistakes by democratic states to failures of democracies that lead to people making a case for authoritarian responses that step in to fix things.
“Taiwan is living proof that you can have both, both liberty and freedom, human rights and democracy and economic growth, public health and so on.”
“One good thing about democracy is that it’s resilient and that whatever it learns is in the memory of the people, not just a few elites that make the decisions. So more democracy, not less, just because democracy didn’t perform that well in one particular pandemic or whatever.”
I’m working on democracytech, how do I engage with Taiwan?
Come and visit, become a resident, you can work for your own company (you don’t have to work for a Taiwanese company), get universal healthcare, be part of the civic tech scene and propose ideas in the presidential hackathons. You could get your ideas made into policy!