We’ll be on the record. People can see it, but only after we edit for 10 days, so we can edit it in private.

How often?

Like monthly?

That’s great.

Yes, that’s the official title.

I do agree.

Yes, I did.

Of course. I did homework.

Explain-ability.

Fair enough. Because the white paper frames these questions around AI but really only the AI explicability is pertinent to solely AI. Everything else applies to non-AI technologies as well.

For example, data ownership and privacy, regardless of whether it’s machine learning, doing data processing, or just a very rule-based engine doing the processing, the same privacy and data ownership predicates apply. These are not, to my view, specifically AI-ish things.

AI does change the equation somewhat, mostly because if it’s a machine assuming agency, it is very difficult from a legal perspective to attribute the accountability fairly. Because the AI essentially is a agent that is currently unable to defend for itself on the court in a normal interrogation or accountability process.

This is not saying that Ai will not develop this capability. This is merely observing the fact that most of the machine learning algorithms that are in use today is not designed with explain-ability as its main design criteria. People essentially have to perform forensics to uncover how its reasoning went or didn’t go after it even happened.

Unlike traditional systems, AI systems, it’s even more difficult to add such explain-ability functions after the fact. For normal rule-based system, sometimes the rules are so byzantine it’s impossible to explain either. Same concerns do apply.

At least because it is componentized in a way that is logical as a system programmer, I know that if I have sufficient instrumentation component to sufficient amount of parts of the system, accurate enough image will emerge. I can trust the flow chart.

If machine learning enters the picture, then we have to interrogate not only the design of the system but also all the data that was previously used to train the system. Essentially, its memory, its context.

That are often either inaccessible or not accessible in a form that can be understood by humans within a human life span or not labeled enough so that even if there is bias, there’s no systemic way to detect such bias. The training phase of machine learning does add to the complication of explain-ability.

I think even for explain-ability, just as other three points, these are general to all computer systems or SAT systems, and not at all particular to AI. AI makes it more complicated, but even without AI they don’t go away. It’s the first answer.

Particular to AI, we have a series of websites that outlines both the action plan and the latest development. For example, for AI, it’s just ai.taiwan.gov.tw.

Right. They’re new as of this year, so OK. Right. [laughs] It’s just ai.taiwan.gov.tw, and it’s one small part of many...

You mean AI Taiwan, you do...

Yeah. If you google for AI Taiwan, I think that’s the first stage.

If you go to the DIGI⁺ Taiwan website...

I think it has the English version, so we have a dgey.gov.tw, which I think it’s this one rank. You probably know that, and this is on the front page, and if you click English, it also goes to DG+ Taiwan, in which AI Taiwan is one of the subsites, but there’s also BioMed, ASP, and the usual suspects.

The whole point of using this domain is stressing the fact that this is a cross-ministerial endeavor.

https://smart.taiwan.gov.tw/.

Yeah. We begin with smart.taiwan.gov.tw., but Smart Taiwan links to AI Taiwan. It also links to BioMed Taiwan, which is the biomedical part. They all look very similar. Underneath it, of course, different ministries maintain different sections, but the reason why we used taiwan.gov.tw is to stress the fact that this is a cross-ministerial administration level policy.

Yeah.

It’s a very useful provocation.

That’s for sure.

There’s many other provocations. The BioMed portal, which is biomed.taiwan.gov.tw, but you can also use BioTaiwan. The fun story is that different ministries prefer different subdomains, so we’re like everybody gets a domain and it’s the same website.

Because the medical technology people think "bio" is a bit research-y, and it doesn’t really include them, and so they want a different subdomain, but it’s the same site. We have to be politically correct even on a shared cross-ministerial domain...

You’re guaranteed to access the same content whether you enter as BioTaiwan or BioMed Taiwan, whether you enter as Smart Taiwan or DIGI Taiwan. It’s all the same website, so [laughs] that being answered.

I think data governance or data agency the main provocation to Taiwan’s administration this year is not AI, it’s GDPR. GDPR is by far the main provocation for us to take a systemic approach reviewing those are existing privacy act, which is loosely modeled after EU anyway -- its previous version anyway -- and also to review the need of having a data protection authority or a DPA.

Without a DPA there’s just so much that law can’t define, and that interpretation is bound to differ widely between different ministries. Even for very simple things, like what is "informed consent", or very simple things like does this belong to sensitive personal info, non-sensitive personal info, or not personal info at all, different ministries have different standards.

That creates a real problem for entrepreneurs, because when they work on things that’s spun small people ministries, often they will ask the local city government for help, but they get different answers depending on which bureau they ask or which department they ask, because each department correspond to a different ministry.

Each ministry has its own different guidelines as of how "informed consent" works, how digital signature of a council works, and so on. This year we adopted a basic idea that there will eventually be a data protection authority. Working in anticipation of the DPA, the National Development Council founded the Office for Personal Data Protection.

The NDC Office has two main mandates. The first is that it’s charged to do adequacy negotiation with the EU, so acting as kind of an internal DPA to negotiate the adequacy constraints. The second thing is that it’s harmonizing the different interpretations of different ministries.

Previously that role is reserved to the Ministry of Justice, but the Ministry of Justice has historically said first, there’s not sufficient human resources, and second, they don’t actually know how it works in the field anyway.

They are more concerned about interpretation of the legal text rather than how people are actually interacting through a screen on the inform click-through, whether the privacy policies are displayed correctly or not. They don’t have that much connection to the industry to tell and to learn about the latest developments on privacy has in technologies.

Maybe the NDC is better to do this, and the NDC is willing to take on this responsibility, so it is officially delegated from the Ministry of Justice to the NDC’s data protection office in an administration meeting and it’s on public record.

We are at the point where during the next two years or so, as part of the adequacy negotiation, the NDC’s data protection office will bootstrap a real DPA whether that will still be a DPA under NDC, whether NDC will extend its mandates to become DPA-ish, whether there will be independent body, we don’t know.

First, that depends on how strong EU is acting and reacting on adequacy constraints. Second, it depends on how quickly the NDC can harmonize the existing interpretations.

It’s cross-functional. Many of them are from the Regulatory Reform Center of the NDC. I think the chambers are pretty familiar with that particular unit. They’re the unit who handles the EU and American Chamber [laughs] white papers and turn them into regulatory compliance parts.

They are also helped by the various like III or ITRI, who have already invested considerably into privacy-enhancing technologies. They will be helped by the technologists, while being at the heart, at the core, a legal tech entity that develops new solutions and new interpretation, but to serve the regulatory process rather than the other way around.

I think that’s pretty fair. It’s also a very good sign if the regulatory units within each ministry is willing to work together to uphold data agency. Then it’s much more likely that a harmonized interpretation will ensue.

We’re also talking about enlisting the help of the statistics functions in each ministry. There is a new law that’s already passed, the new statistics law that defines data interoperability between government agencies, and even across ministries, for statistical purposes.

Within the Data Protection Act, there is already a clause that says if it’s in aggregate, if it’s statistics, if it’s for the common good, then it doesn’t need to be treated as personal data when using it.

It’s still, of course, personal data when processing it. When you run a statistics function, its input is personal data, but its output is no longer. The person who can run this is still only the data operator. They can’t just randomly assign somebody else to run this data processing for it.

We’re sticking to the term the law uses, which is statistics, so making it into a form for statistical use or statistical consumption. De-identification is kind of an unfortunate term, because it’s not in the law itself.

The law is written with an EU context, but de-identification is thoroughly a US term.

It could be taken to mean anonymization in a GDPR context, but sometimes it’s also taken just to mean pseudonymization in GDPR terms, like exchanging it to a pseudonym that could be linked back afterwards. The term itself is ambiguous. Some people see "de-identification" just as masking.

Masking the ID. Sometimes people see de-identification as being total anonymity.

The term itself being ambiguous, it’s not the most helpful term for us to use, which is why we’re gradually phasing out its use. If you look at the new statistics law, and the accompanying regulation to the new statistics act, you will see only personal data or big data processed for statistics use. That’s the term that we’re using in the statistics regulation.

It removes ambiguity. Basically, if there is any chance of re-identifying individuals, nobody will accept it as statistics. It needs to be in the aggregate. That’s what statistics mean. The new statistics law gives each agency the possibility, and indeed the right, after processing the private data into a statistics form, to exchange freely across the statistics agencies.

One ministry, when it want to ask a question about the population or about epidemic, about public health, whatever, it doesn’t need to redo the data gathering. It can just query the statistics form of each other’s ministries. This is, for me, much more clear-cut than the previous, for research purpose and for public good, which is so ambiguous you need to be evaluated thoroughly on a case-by-case basis.

Now the statistics form could be independently verified. This opens the door to a global movement called Open Algorithm, which is the scientists and researchers publishing better statistics algorithms that takes personal data and outputs something that is more useful than the traditional ways of doing statistics.

It’s still doing aggregation. It’s still doing statistics, but it’s doing it in a way that’s very useful and could be comparable across jurisdictions. Whereas before, each agency do their own statistics, but they do it in different way. It was different parameters.

The end result is not comparable. If you try to feed AI with these things, the AI develops bias that are impossible to detect because the bias are inherent in the way the data was processed by different jurisdictions in the first place. That’s a nightmare.

What we’re now doing is we can publish the schema, that is to say the shape of the data. We maybe publish synthetic datasets, that is to say it doesn’t really contain any personal info because they’re all fake. Given those random data, the data scientist can do a lot meaningful statistic algorithms, but still for the data operate your statistic agency to run and then to publish. Then, it could be comparable across jurisdictions.

Yeah, we have a website for that too.

Yeah, but it’s not yet completely translated to English. It’s CI Taiwan, so not AI, but CI, CI stands for IoT, but it could also stand for collective intelligence very quickly. We’ll also have a SI Taiwan for social innovation soon. In any case...

CI Taiwan answers your question by essentially unifying the data contribution and data collection efforts from the civil society and the private sector. For example, for air quality, there is a very famous community called LASS or Air Box that are part of the Taipei City’s environmental education in primary schools that collects by themselves entirely, without the government or intervention.

Here, you can see very quickly how the air is like across Taiwan and as contributed by 2,000 or so independent measurement units. The cool thing about this effort is that it doesn’t use any environmental protection agencies’ machine or anything like that, but rather it is entirely what we call citizen science or crowdsourced.

It also is somewhat unique in Asia because it depends on the complete freedom of assembly and speech. In other jurisdictions, it doesn’t take 2,000, maybe just 20 or 200. People are doing this and the government will feel its legitimacy is being threatened and it will try to shut them down.

We don’t shut anything down. We instead join them. We say they’re already covering many points that are close to people, but for more faraway places, places where there are gaps and so on, the government will contribute those gathering.

That’s right, complementary. Then, we’re also working with E3 to manufacture cheap, but still more precise PM2.5 sensors, so they could be upgraded to be more trustworthy. The same community is now doing water quality measurement as well.

Basically, the government, as part of the forward-looking infrastructure plan, allocated NT$4.9 billion over five years to build the CI project, which is essentially saying all this data from different heterogeneous sources are all stored in a national center for high speed computing, which we bought a lot of NVIDIA GPUs.

In the future, any professor who wants to do analysis emission learning, they can just upload their algorithm here, instead of having to download the datasets from the diverse sources.

At the end of this year, we’re having the first public challenge, where we challenge the people to use those streamlined data contributed by all those different ministries and civil society to develop meaningful social impact applications, and to create a data economy out of the shared data based on international standards, such as SensorsThings API and many other APIs.

This is not Taiwan only. We’re harmonizing all these data formats to be compatible with global Geospatial Consortium standards. It could conceivably be bridged to all the different regional sources as well. It’s not specific to Taiwan.

What it does is that it shows a model where the user contributed or citizen contributed data can enter part of the policymaking process and that we can vet each other’s accuracy, not by writing essays on major newspaper, but rather comparing data in a meaningful way.

There’s already people working on distributed ledgers that are taking checksums or stamp shots of this data and storing them on mutual distributed ledgers. People will have the confidence that we’re not changing the numbers the day before the election, just make certain...

Right. The data pipeline is threefold. The first thing is the raw data. The raw data just gets entered as is. Then, the private community themselves have a mechanism, like the Air Box community, they have a mechanism to detect off-the-chart or unusual points. So many of their sensors are not very precise, to begin with.

They have developed a very sophisticated self-calibrating algorithm on the community side. Then, instead of just saying we are completely complimentary, of course we still have sensors on duplicated sites built by the government and also by the civil society just to cross-check each other.

On the third stage, we have, during the data aggregation phase, a way to compare the long-term trends of one particular contribution from one particular community versus another cohort of sensors. This is not to detect malice, but more likely to uncover the biases inherent in the kind of sensors that they use.

Maybe in the hardware, above a certain range, it starts to become erratic or below a certain range it becomes deteriorated. It’s possible. Basically, the first stage is to look for algorithms that can systematically uncover and correct bias.

That, in fact, one of the points that we’re looking for in the Grand Challenge at the end of this year is to get the researchers a set of data and develop a way that could increase the net trustworthiness of the diverse data sets. That, too, is part of the challenge.

It will all be announced on CI Taiwan. We choose this in particular because there’s no privacy issue. We’re not saying that it cannot one day be applied to, for example, health care data that are donated by home caring stations or by the patients themselves.

Basically, the privacy, the DPA adequacy, that is one arm.

Statistics is one arm. The environmental data aggregation, citizens contribute the data. We’re running three consultations at the moment about interchange of data across ministries, publishing of government data to our citizens, and how citizens-originating data enters into policy-making process, all the three directions, as part of the vTaiwan process.

Just this morning we ran the second consultation on government data publishing data outward. Soon, I think a week or two, the third one, the incoming one will...It’s all live-streamed. If you go to the vTaiwan Facebook group, it’s all live-streamed.

We will aggregate all these recommendations and figure out whether we need a act-level change, a law-level change, whether the law itself is sufficient. For DPA, of course we will have to change the privacy law at least at one point, the current authority chapter, and maybe data localization chapter.

Aside than that, maybe it’s possible to do it entirely on the regulatory level and not on the law level. This consultation is mainly to determine whether there’s any input from the community that is only realizable by changing, say, the Freedom of Information act or the Privacy Act. If not, then it’s all going to be regulatory level. That’s the legal part.

Then the CI, the civil IoT, is progressing without worrying about privacy. Basically, we build the standards, the technical expertise, the trust between the civil society, private sector vendors of those sensors, and the ministries involved to try to convince the minister it really generates better social impact if we work together.

Once these two are independently solved, then we can tackle the most difficult part, which is the use of private data under consent, for public good.

Right.

First, I’ll encourage you to review the DPA discussion on vTaiwan where it also touched on the localization. The main provisional idea is that it all depends on how strong the DPA is.

If the DPA is a thoroughly EU-style DPA that has de factor administrative not just interpretation control, that is to say it can recommend specific privacy-enhancing technologies, if it has this level of control, then, of course, a EU-style exchange only with adequacy countries or jurisdictions makes sense.

If, at the end of it, we pass adequacy, but with a DPA that are more advisory than prescribing particular PETs, then maybe it does make sense to keep the current data localization idea, which is open by default and closed only for security processes or whatever. We will know when the actual DPA gets bootstrapped.

Depending on the strength of the DPA, we will know the strengths of data localization. Personally, I don’t have a preference. I think the consensus now is more of having the citizen being aware that they have certain agency under not just GDPR, but existing PIDA. They can demand explanation. Data is not just an asset. It is a relationship.

If sufficient amount of people in a civil society takes this view, then maybe we don’t need a strong DPA. Maybe the citizens, as part of consumer protection associates, authorities, and so on, or like iWIN, is able to safeguard it to a certain degree. That would actually be a pretty balanced power between the civil society, the private sector, and the government.

On the other hand, if the civil society does not have this kind of empowerment, we don’t succeed in getting people taking the view that they have agency, that data is a relationship, then a much stronger case will be made in the Parliament about a strong DPA. That’s just logical, and I don’t have a preference one way or another.

Yeah.

It’s not always like that. During the Company Act rewrite recently, we explicitly said the US approach works better in many things, so we just adopt this.

I think the mindset is changing. One part of it, I think, is the successful application of a doctrine of the so-called sandbox laws. It is not a legal term. Technically, it’s experimentation and advancement of economy or whatever. I don’t know the full name, but everybody just call it sandbox, so might has well. [laughs] At the moment, we have three sandboxes.

The regulatory-level sandbox as offered by the SME agency, the regulatory reform center’s sandbox. They don’t call themselves a sandbox, though. They are at https://law.ndc.gov.tw/ which is a platform economy regulatory adjustment, where the NDC, the Regulatory Reform Center, the RRC, takes people’s appeal that certain interpretations are out of data and work with the ministries to abolish them very simply put.

The SMEA sandbox is more about a new emerging trade or a new emerging business model. It’s not even sure whether it runs counter to our regulation or not. It would help to clear the gray area and to even work out new regulations to fit their purpose, instead of retrofitting existing obsolete regulations.

The SME one is for when you don’t know what’s blocking you. The NDC one is for that you know precisely what need to be abolished in order for you to move forward. They have this relationship.

Sometimes the SME one passes to the NDC. There’s a recent case about Upark, a startup that installs this kind of blocking thing on the parking lots.

The SME worked with NDC to relax that, so that now on average, if you just rent out up to eight hours per day, throughout the month, then you are not taxed and hold like a operating parking lot. That is a very good case of the SME taking it, clarifying it, passing it to NDC and NDC helping it to adjust the law.

This is much more...

It really worked. Then, this is less continental law system. This is much more US style and much more case law-ish thinking. Even if its challenge is lost, we have the FinTech Sandbox, which admittedly suffers from the fact that many people entering the FinTech Sandbox is not even very clear which law they are challenging or which part of the law that they’re challenging.

There’s a lot of handholding to do by the FSC. If you listen to some other side of people, they say they have a lot handholding to do to the FSC people, so both sides, sure, have a lot of handholding to do, but as a FinTech.

We are also working on the AV, the autonomous vehicles sandbox. We expect that to pass this session or latest next, but likely this.

The MOEA will take the application and do the evaluation. When it comes to making exceptions on existing loss, the MOTC will do the exemption. All the while, the evaluation of the safety, accountability, AI applicability will be contributed by MOST.

Verily, but the alternative is to have the MOTC make all the decisions itself...

Yeah, you go to your local MOEA office.

Yeah, the municipal MOEA office.

Well the MOEA has so-called 馬上辦 service centers. They report directly to the MOEA. It just happens that they have multiple municipal offices.

Yes.

The process, according to the draft law is all online, so I’ll just explain very briefly.

Basically, the MOEA is the driver, just like in the SME sandbox. The applicant, I think Upark folks, doesn’t even need to know that there are MOTC, MOF, NDC, RRC, whatever, involved. At the beginning, it just talks to the SME folks in MOEA.

Yes.

Yes.

Yes.

Yes, for the AV sandbox. The AV sandbox it’s also paramount for us, for all the ministries to be in a multi-stakeholder panel because they all have some stake.

The panel need to make decisions in a way that is not detrimental to any ministries’ purposes. The MOEA is better at handling this kind of innovators proposals because it’s more flexible in understanding the language that are not overtly technically or legally correct.

The thing with FinTech Sandbox was that there’s a lot of language difference between the innovators who describe the latest immutable distributed ledgers and the existing bankers describe their compliance roles. They are literally worlds apart.

The FSC devoted a lot of time just to reconcile their languages. In the time for the AV sandbox, we think it’s maybe better to have the people who work with startups all the time, the MOEA people, to serve as their face, their proxy, and then talk to the ministries that are really needed to relax the laws or their existing laws.

Now, the municipal government is still important, not in the sense that they have a veto on the roads or whatever, but rather publishing the real, actual social needs of the city government.

For many AV experiments, it will fare much better if you know exactly what, for example...I’m just making this up, the Penghu County. They really want smart ships because for some offshore islands, the transportation situation’s very difficult. They cannot keep the sailors around all the time.

If there are some lightweight, self-driving ship that people can summon at will, it would really help them. They have all the relevant climate data, sea data, whatever data to support the parameters on which to make such experiments.

Instead of starting from zero, the innovators can just look at each county and region’s needs and not compete, but rather try to solve a little bit of it in a smaller section. The relation is not like innovators are in the dark and people just randomly veto them, but rather the different municipalities declare their needs.

We’ll publish this as a transcript.

Yeah, I’ve been thinking about it a lot. One of the first thing that we can do is simply to make these websites Google Translate, frankly, and even in English, where it makes sense. Our communication so far it’s maybe only...The English version of these websites is maybe only two percent of the information.

Yeah, it’s horrible.

Google Translate does a better job.

Yeah, it’s really, really bad, which is why we asked for the Taiwan.gov.tw domain name because we think for a foreign person, it’s impossible for them to know that for the SME Sandbox, you have to go to SME.gov.tw.

For the technology one, you have to go to MOST. Yeah, it’s important.

Which is why we have this new generation of website that all ends in Tawian.gov.tw and it’s just a better English face of all the nationwide activities that we’re doing, and the index of which is SmartTaiwan.gov.tw.

We’re still working hard with the Ministry of Foreign Affairs, who gave us this domain name, how to unify the strategy of our new generation of websites versus the existing one, which is Taiwan GOVTW, with the ultimate goal of this being redesigned or shaped into something that can tailor to the different needs.

No, we can edit out.

What do you feel about Contact Taiwan?

Contact Taiwan.

People don’t...

This is Contact Taiwan.

It is also part of it.

Like this one.

You do mean this website?

I do...

I do agree, like totally.

I do agree. I haven’t seen this website before.

I usually recommend to people Contact TAIWAN. I do see that there are sister sites somehow.

Maybe e-Taiwanese. [laughs]

There’s a gathering by Tricky Taipaei, that there’s a community of people who work in marcom and are brainstorming essentially exactly the case you were just making, like how to make the message about Taiwan consistent.

This is today’s talk.

The three panelists are Didi Bethurum, former global marketing director at Gogoro, someone from the Tourism Destination, marketing consultant, someone from Bloomberg, a travel writer/journalist, and also Kathy of Tricky Taipei.

Tricky Taipei is this website that offers some very constructive criticism of Taiwan’s message, not just about tourists, but also about the English presence of Taiwan’s websites and things like that.

There’s a community already forming about how to improve the message. My colleague in PDIS, in my office, who’s from the Ministry of Foreign Affairs, will attend. We’re trying to build a better relationship to not just these communities, but also maybe Forumosa and other only communities.

Or rather on the lifestyle part.

Is the investment and business part.

I’m aware of that. The problem, as I see it, is that in those two angles, the existing English websites, they often confuse the two messages. You see the Startup Taiwan or Contact Taiwan talking about the food and hospitality...

You see the communications tourism bureau talking about "five plus two" industrial innovation. It’s very confusing to both audiences, which is why I do think that delineating and making clear messages on one or the other will make it much more clear to the communication crew in our ministries how to position to different target audiences. Currently, it’s very confusing.

Sure.

Has to adjust, of course. I do agree. I’m just saying is all we have at the moment is taiwan.gov.tw, the domain name, the goodwill of the MOFA people to help reconstructing the message, some links to existing communities, and a few new initiatives like AI Taiwan, Smart Taiwan, CI Taiwan, Bio Taiwan, that we can do it our way.

At the moment, we don’t have a systematic way to rein in existing, maybe higher Google-ranked, domain names that are of previous administrations out there. I do have some thoughts about it. Every year, we hire anywhere from 25 to 40 interns to look systematically to all the different government websites, like 500 different websites.

The first year, last year, we checked compatibility with iPad and Mac. This year, we check compatibility with iPhone and Android phones, so that they don’t feel overly desktop-y when using it. They’re very helpful.

I’m thinking maybe early next year, our third batch will be just hiring people with good English abilities to look through all the website that proposes itself to target a English audience and just systematically remove anything that is actually detrimental or replacing it with something that feels better.

We can only do this if we have a more consistent message criteria upon with the interns can help look into existing ones and evaluate it according to that criteria.

If you’re willing to coach the students, I’ll let you know.

They have multiple phone lines. You can just call and they come to your place.

Yeah, because they serve a non-fiscal purpose.

The whole point of doing social innovation, the National Social Innovation Act and the plan is to let the people working on environment or social values to see economic values as not something that’s detrimental to them, but can add to them.

It starts as the Social Enterprise Plan under Minister Feng Yen, but now I’ve expanded as the Social Innovation Plan, part of which is the Social Innovation Lab here. The point of that plan is I tour around Taiwan, I meet with innovators, who are doing business models, but are also solving real social needs, fulfilling real social needs.

I tour around Taiwan. In the past few days, I went to Miao-Li and through teleconference, the 12 different ministries, here in the Social Innovation Lab, see through my eyes essentially what the local innovators are doing.

Instead of just segmenting into abstract words, they can see that it is possible to work at once to further the economic and social and environmental goals. That new generations of entrepreneurs, during the Digital Innovation Forum, Gogoro is a very good example or PaGamO or whatever.

They start with a clear social mission. They can even name the sustainable development goal that they’re achieving, but while building a very good case of a supply chain, a business model, and so on.

It is only through exposing the public servants in the Ministry of Education or Health and Welfare to those successful entrepreneurs first hand, using their services or at least watching me using their services, they can understand that it is not there to create more headaches for them.

But rather, if they allow flexibility, it is ultimately better for their own purpose as well. It’s through these tours that I encounter. For example, there’s a Christian association, the Good Shepherd Church Association.

They’re very successful in giving the people who are of old age out of their houses and to have lunch together and develop a lot of relationships with nearby kitchens to turn almost waste into useful Bento Boxes and whatever.

All the while, they also develop some successful product lines of drinks, also overseas exchange programs and whatever that are also profitable. They’re now looking to move the parts that are profitable into a company, while the part that are charity are still kept as the association, but have the association control the company through what we called the closely held company law.

They can have special voting rights and they can even hold stock without having money contribution, by essentially having the work count as the initial stock in funding the company.

Now, all this is technically legal, but nobody really wants to be the person to approve because they are afraid that it will go to unmanageable places. I personally chaired one of the SME Sandbox cases also, the cross-ministerial meetings that basically said, "We’re going to hold this company under high scrutiny.

They have to declare their social mission in their funding documents to align with the Controlling Association." They need to do public reporting together, and so on. Satisfying these agreements, we’re now comfortable with an association holding essentially become a stock holding, controlling entity of its underlying corporation.

This is done. Then, many other charities are now working toward similar goals. The MOHW, after watching this for a while, now says, "OK, so foundations under MOHW are now also qualified to control subsidiary corporations in exactly the same the same way if they fulfill the same accountability purposes."

If not for the face-to-face consultation and meeting with the people who are actually doing the work, they only think in abstract ways and they only think of the drawbacks and loopholes. But having a fully transparent record of them raising those points is essentially me absorbing all the risk.

If it works, everybody shares the credit because you can Google and find their names after various different ministries. It worked pretty well, and now we’re also working on similar cases with clubs. That’s another entity type.

This is the long answer, but the only way is to let the Ministry of Education or of Health and Welfare see that the fiscal and economic drives are good ambitions. They are ambitions that helps them fulfill their social duties. So far, it’s been working pretty well.

Yes.

Yes, there is a fund for that. Also, because we’re now calling them social innovation companies, they automatically qualify for innovation related loans. They qualify for loans which requires innovation, but having social innovation as part of innovation, it helps, and to be listed on TPEX’s GISA.

They also qualify for the NDC NGO fund. The NDC has several funds, but having social innovation automatically qualifies as a innovator if innovation is one of the criteria. So it’s as good as, for example, winning an innovation award.

I’m here every Wednesday, so feel free to drop by anytime.

The teams.

That’s great. We’ll definitely be in touch.

Thank you.

We’ll definitely be in touch.

That’s right.

That’s great.

Take your time.

Nice to meet you.

I will send you a transcript...

Thank you.