The Additional Lessons Should Be Taken In The COVID-19 Pandemic, except for Bill Gates’s Suggestions – (5)

This is Part 5 of this article. Read the last part here: Part 4

6. Recording The Country Population

Bill Gates points out in his book that some countries don’t do registries of births and deaths to record their population precisely. He said, “Many of them (low-and middle-income countries) estimate the number of births and deaths using household surveys that are held several years apart, which means they don’t have precise data”. Further, he states, “At a minimum, many low- and middle-income countries need stronger registries of births and deaths, information that would feed into national disease surveillance work”.

Indeed, for example, China has had a serious problem with recording birth and death. According to the research from Renmin University of China, the death rate in 2020 in China is 10.07%, and the birth rate is 18.38%. Both figures are higher than the rates of 2019, which is at odds with the declining birth and death rates in China. The research points out that the above serious deviations are possible due to the differences in the data collection methods of the Quipu, which is that there are a lot of suspicions of under-reports in the population data collected with methods of the Quipu. The research also points out the same problem as Bill Gates described – The number of China’s population, birth, and death comes from a sample survey and estimation based on that survey, which makes the number population inaccurate, and makes it possible that the government deliberately under-report the number of births ^[1]https://zh.wikipedia.org/zh-tw/中华人民共和国第七次全国人口普查  ^[2]https://web.archive.org/web/20211101150221/http://pdsc.ruc.edu.cn/jdjx/55f514b5aa1a4f6397f535dc762438b0.htm#:~:text=由此推算2020年,出生率、死亡率严重不符%E3%80%82 .

This problem must be solved. I suggest that the Chinese government not only start to do proper registries of births and deaths but also advance the registries to an extent that is far beyond the current registries many countries do. For example, both the U.S and Taiwan require a birth certificate to register a newborn ^[3]https://www.parenting.com.tw/article/5087398  ^[4]https://www.usbirthcertificates.com/articles/birth-certificate-baby . China does that, too ^[5]https://baike.baidu.com/item/户口登记/2492270 . However, there have been some infant/children trafficking events happened in China, which are about human traffickers selling infants/children they abduct to Chinese households who wish to have children but don’t ^[6]https://www.hk01.com/即時中國/649746/失孤-原型尋子24年走逾百萬里終團聚-劉德華拍片祝賀 . It’s possible that the requirement for a birth certificate for the registration of a newborn isn’t executed properly. If the requirement of a birth certificate is executed properly, from my view, it’s possible that the Chinese households who bought those children successfully used fake birth certificates to do a birth registration, which is a case of overreporting. Although as a whole the number of trafficking events in China may not affect the discrepancy between the 2020 birth/death rate and the 2019 birth/death rate – it’s likely that there are both under-report cases and overstatement (overreporting) cases, but the number of under-report populations surpasses the number of overstated population, trafficking events in China is one of the consequences that birth registration (or household registration) in China isn’t done properly.

So how to resolve this issue? Firstly, the Chinese government must identify what the problem is. There is no explanation of how a household registers the baby they buy – probably because no one thinks of it when reading about a child trafficking event; it’s still possible that the requirement of a birth certificate isn’t executed properly. If that’s the case, just execute it properly – many heartbreaking parents will be able to get their children back, and part of the restatement (overreporting) issues can be solved. Nevertheless, if the problem is that households use fake birth certificates to register the babies they buy, I suggest adding anti-counterfeiting features on birth certificates, just like the anti-counterfeiting features on an identity card. Anti-counterfeiting features don’t exist in birth certificates in both the U.S and Taiwan. Add such features is advancing the registries of birth far beyond many countries do. I also suggest the Chinese government consider making birth certificates more advanced, for example, making birth certificates digital and accompanied with chips. This way can not only prevent fake certificates but also provide other benefits – adding a digital DNA record to the digital birth certificate, for example.

Making birth certificates digital with anti-counterfeiting features will deter many households who want to buy a child from actually buying a child – if they can’t register their children and make them able to go to school, raising a child will be partly meaningless. If the registered problem of children who are trafficked is solved, not only the figure of the birth population can be more accurate, many heartbreaking events of losing children won’t happen again.

7. Digital Platform Of Pandemic Matters For Civilians

Bill Gates discusses several problems related to digital platforms of pandemic matters in his book as briefly described in Point (1) to (3) below. I will add some additional issues in Point (3) and provide my suggestions for all three points including additional issues.

7.(1) Language barriers:

Bill Gates: “Setting up a platform which makes the test, creates the website where people can sign up and processes their samples is only part of the challenge…It is another matter altogether to make sure that the results reflect the actual makeup of the community. Not everyone can navigate a website easily. Language barriers can get in the way. When demand for test kits is high and supplies are limited, people who can stay at home repeatedly checking a website have an advantage over essential employees who still have to go to work.”

7.(2) Putting regulatory submissions in the cloud and a standard format

Bill Gates: “Another innovation I hope to see is putting regulatory submissions in the cloud and a standard format so that they can be reviewed by all the regulatory agencies across the globe without duplication. And in the United States especially, adopting a standard format for patients’ health records would have many benefits, including making it easier to find potential volunteers for drug trials.”

7.(3) Outdated software

Bill Gates: “Public health institutions don’t get as much public attention or government funding as they deserve—not at the state level (including in the U.S.), national level, or global level with the WHO.” He further says, “In 2021, Microsoft worked with one U.S. state health department whose software was two decades old.”

Taiwan also has a similar problem. At first, I thought it was a problem that only happens in Taiwan; for a highly-advanced country such as the U.S, the software it gives to health professionals and normal civilians must be up-to-date. It was until reading Bill Gates’s book, I started to realize the problem exists in the U.S, too. Anyway, let’s also talk about what Taiwan’s problem is. A legislator, Ms.Zheng, criticized the platform from our IT governor, Audrey Tang, who is a respected programmer providing consultation service to Apple according to Wikipedia. Ms. Zheng criticized, “Every platform made by our IT genius is slow, and there is always a problem when the platform goes live. Any ordinary engineer can make a platform that is a hundred times better than what Audrey Tang made. ” The problem when the platform went live is about a platform of registering vaccine willingness went crashed when there were 2.1 million people online ^[7]https://www.cna.com.tw/news/firstnews/202107135009.aspx 

Which side is right? The opposite side that objects to Ms.Zheng’s criticism of Audrey Tang’s platform seem reasonable. However, I agree with Ms.Zheng’s opinion; the platform sucks. Let me explain why.

7.(3).1 Traffic

A person who agreed with Mr.Zheng’s criticism said, “It’s just a high-school level system. Wonder how many dollars from taxpayers are wasted. ” Someone replies, “Are you able to find a high school student to develop a system that can deal with 10,000 requests per second? ^[10]https://today.line.me/tw/v2/comment/article/wD1gXE ”. Well, I suspect the traffic was not 10,000 requests per second. Let’s take a look at the actual figures. In one event that the platform crashed, it’s 6000 people per minute; that’s equal to 100 requests per second. Because the platform crashed, the news headline is “The platform is in a heavy traffic jam! Over 6,000 people are online per minute. ^[11]https://www.ettoday.net/amp/amp_news.php7?news_id=2029533 ” In another event that the platform crashed, it’s 30,000 people per minute; that’s equal to 500 requests per second ^[12]https://www.setn.com/news.aspx?newsid=953064. Both are not 10,000 requests per second. I suspect the person who said 10,000 requests didn’t actually calculate it or doesn’t even know how to calculate it.

Is dealing with 500 or 100 requests per second difficult? A Taiwanese complained that the system sucks because it should be able to deal with such traffics. Another Taiwanese who calculated and got a result that the traffic is 22,222 people per minute, which is equal to 370 requests per second, retort the person who complained by saying, “You must have never developed a website? You think it’s easy to make this system? There are 22,222 people online per minute on average (370/per sec).” Another person replied, “If the platform can’t deal with that much traffic, why do they let this amount of people use the platform at once?”. The person who complained provided another argument to refute the argument of dealing with such traffic is “not” easy by pointing out, “550,000 concert tickets of Mayday (a Taiwan rock band) was sold out in 9 minutes. Average the traffic is 61,111 requests per minute (1,018 requests per second)”^[13]https://www.setn.com/m/ampnews.aspx?NewsID=610044  ^[14]https://www.dcard.tw/f/trending/p/236475848 . I agree with this person’s view – The developers of a concert ticket platform can make the platform be able to deal with much higher traffic. The vaccine platform from the government can’t so it sucks.

In addition, regarding the point that the government shouldn’t let this amount of people use the platform at once, many Taiwanese agreed with this point and said the problem is because the developers of the vaccine platform didn’t do a stress test and calculate accurately about the possible volume. The engineering department of the company which develop the platform also stated that it’s because their estimation of traffic was too low, so they didn’t buy enough servers. Another root cause in other events is that the traffic exceeded the load of the front-end gateway^[15]https://www.cna.com.tw/news/firstnews/202107135009.aspx . Nevertheless, in my view, this is not a good reason.

First, which is a known fact. The platform crashed many times, not only one time. If you didn’t pay attention to calculate the potential traffic or didn’t know how to calculate the potential traffic in the first time, and you didn’t prevent a crash by reviewing all possible aspects in the second time in July 2021, why a crash still happened in Jan 2022 ^[16]https://www.ettoday.net/news/20220115/2169708.htm ? I think there is a big problem with the ability of the development vendor. 

Second, the order of inoculation was already defined and went public at that time, how many population in Taiwan is a known fact, and how many population in Taiwan is qualified in a specific order is a known fact that can be calculated. An accurate estimation shouldn’t be that difficult, especially since the traffic was only 500 requests per second. 

Third, this point will be a bit technical, but please bear with me. You can understand the later part and don’t have to understand the full part. Ok, let’s start. The first time the development vendor said they didn’t buy enough servers. The second time the vendor said the servers are enough, but the traffic exceeded the load of the front-end gateway. I am not sure, but I assume “the load of the front-end gateway” means a front-end load balancer. Here a question comes: It’s load balancers or instances which couldn’t handle the traffic? Anyway, it doesn’t matter that either it’s load balancers or instances. Either a load balancer or an instance can be charged based on the actual usage or have multiple choices based on CPU and bandwidth for handling corresponding traffic^[17]https://aws.amazon.com/elasticloadbalancing/?nc2=type_a^[18]https://aws.amazon.com/ec2/instance-types/ . Now comes the non-technical but important part. The first time is that didn’t buy enough servers; the second time is that didn’t buy enough load balancers^[19]https://www.sumologic.com/insight/aws-elastic-load-balancers-classic-vs-application/  / bought the wrong type of the load balancer / chose the wrong type of instance which can’t handle high traffic. It’s likely that the vendor “treat the head when the headaches; treat the foot when the foot aches (a Chinese proverb expressing it’s a piecemeal approach)”, and didn’t have an experienced and comprehensive review on how to prevent crashes when the potential traffic is known.

Although the evidence of Mayday’s concert record is clear, the person who thinks the system is “not” sucks still didn’t believe it and stuck to his position. In fact, there are already many people saying “Concert tickets, Taobao platform during Double 11 festival…etc; these all experienced a high volume of traffic and didn’t crash”; however, many people didn’t believe that the platform sucks. Why? Because they didn’t provide actual data of the example they listed, like the Mayday example. Many person’s arguments with examples weren’t presented with actual data. If it were me, I would wonder, “Is the traffic of a concert ticket platform or Taobao during the Double 11 festival indeed higher than the traffic of the vaccine platform?” That remains to be validated before the data are presented. Nevertheless, the person who didn’t believe that the platform sucks even when the evidence of Mayday’s concert record is wrong – since the data is presented, he should believe it.

Still, for people who don’t believe the platform sucks even when the data is presented, let me provide another perspective. Even if there is no data from another platform, it’s still suspectable the claim that dealing with such traffic is not easy. Is dealing with 500 or 100 requests per second difficult? I am not a software engineer, but 194 requests per second seem low and should be not a big deal for a professional engineer in my view. Even the figure of 500 requests per second should be not a big deal for a professional engineer, in my view – not to mention the pathetic 100 requests per second, but the news ran the headline “a heavy traffic jam!” and many engineers agreed, including a person graduated from National Taiwan University (The top school in Taiwan). It looks like traffic of 100 requests per second is difficult for some Taiwanese engineers. Anyway, if you don’t believe me, the evidence of the 1,167-requests-per-second record of Mayday’s concert tickets is sufficient to prove that dealing with 100-requests-per-second should not be a big deal for a professional engineer. In fact, traffic of 1000+ requests per second is essential for many platforms. 65,000 concert tickets for Adele in the U.K was sold out in 1 minute, which is equal to 1,083 requests per second ^[20]https://metro.co.uk/2021/10/30/adele-fans-gutted-as-bst-hyde-park-concert-tickets-sell-out-in-minutes-15512995/  ^[21]https://www.billboard.com/music/music-news/adele-bst-hyde-park-festival-performance-1235110245/  ^[22]https://www.thesun.co.uk/tvandshowbiz/16581318/adele-tickets-british-summer-time-hyde-park-sold-out/  According to List of fastest-selling product, EXO in Korea even sold 66,000 tickets in 0.2 seconds, which is 330,000 requests per second ^[23]https://en.wikipedia.org/wiki/Talk%3AList_of_fastest-selling_products. In China’s Double 11 festival, Alibaba received 583,000 orders in one second, which means 583,000 requests per second. Google processes 40,000 searches (requests) per seconds ^[24]https://www.quora.com/Google-processes-40k-searches-per-second-On-average-a-web-server-can-handle-1000-requests-per-second-Does-that-mean-Google-can-run-using-only-40-web-servers . Twitter had 70,000 API calls per second ^[25]https://techcrunch.com/2010/09/17/twitter-seeing-6-billion-api-calls-per-day-70k-per-second/ . On the contrary, we were pathetically arguing whether 100 requests per second is difficult ^[26]https://www.bnext.com.tw/article/60038/double-11-tmall-alibaba-2020 .

You say it’s because they are big tech companies? Let me tell you why this reason isn’t working.

First, the vaccine appointment platform is a platform from the government. A government’s work should be better than that most of small software companies. Unless we are in a poor country, the government should have the money required to hire good engineers and pay enough cost of servers required – and we did have; we have 200 million(TWD) ^[27]https://www.youtube.com/watch?v=rDbbxhsHiIA .

Just a note, many people would think that maybe 200 million (TWD) are indeed not enough for such a platform; maybe the platform requires a huge amount of human hours from engineers to complete.

Ok. Two points from my perspective:

Firstly, from my personal work experience as I once worked for a small software vendor company, some small software vendors require only millions of dollars (Unit: TWD) to make software from planning and designing the interface to the development of software, and the interfaces from some software vendors including the one I worked for are more beautiful. The price doesn’t include the costs of servers and maintenance, but even if taking the costs of servers and maintenance into account, 200 million should be sufficient. Anyway, this point is simply judged from my personal experience. Maybe many more software vendors in Taiwan can speak out about whether 200 million is enough for them to make such a platform including design and integration with hospital platforms.

Secondly, according to builtin, the average salary per year of a software engineer in the U.S is $120,039 (USD), which is 3,721,209 (TWD). If we hire 15 engineers for a year, the total cost of engineers is 55,818,135 (TWD), which is about only 28% of the 200 million budget. Given that the main costs of developing software are only human resources and cloud fees such as servers, even if the vaccination platform is much more complicated and requires a higher cost of servers, 200 million should be enough to make a proper website that can deal with at least 1000+ requests per second with enough servers, while most importantly, leaving enough money for a boss to earn profit for himself. But the platform can’t deal with such traffic; the vaccination platform isn’t even integrated with many hospital platforms. In my view, the price of 200 million is supposed to include the job of integrating with many hospital platforms, but it didn’t.

You say that only engineers from big tech companies such as Google are capable of doing this job and their salary is higher than the average salary of normal engineers? That’s an excuse: Point 1, I already lower my requirement to traffic of 1000+ requests per second, not 40,000 or 70,000 requests per sec Google or Twitter is dealing with (Actually, most of them can deal with a much higher volume of traffic). What’s more, integration with external systems is a skill that many average U.S engineers possess. If you can’t do that and you are a software engineer, that’s your problem. Point 2, according to Glassdoor, the average salary per year of a software engineer in Google in the U.S is $205,547 (USD), which is 6,371,957 (TWD). Even if you need to hire Google-level engineers, the total cost of hiring 15 Google engineers is 95,579,395 (TWD), which is about 50% of the 200 million budget. I don’t know. 100 million (50%) should be sufficient to buy enough servers to deal with a traffic of at least 1000+ requests per second with at least little money left for a boss to earn some profit for himself.

I hope these data are enough for you to know that 200 million (TWD) is indeed enough to have a platform that can deal with traffic of 1000+ requests even without considering the first point of my personal experience.

Second, the second point of why the argument that only big tech companies’ engineers are capable of dealing with high traffic doesn’t work is that I think Taiwan probably does have some good engineers. Even if the average level of Taiwanese engineers can’t compete with Google or Facebook, they should “at least” be able to deal with a traffic of 1000+ requests per second. If most Taiwanese engineers can’t deal with such traffic, that’s the problem that most Taiwanese engineers need to improve their capabilities because many engineers in western companies are able to deal with much higher traffic, and the vaccination platform still sucks.  I don’t know if it’s the case that most Taiwanese engineers can’t deal with such traffic, but one thing for sure is that the platform sucks.

*Note: Server cost calculation – Judge based on this example, the cost of handling 23 million users per year is $1,283,400 (TWD), which accounts for only 0.6% of the 200 million budget. Although the situation of every platform is different, server cost with the requirement of the traffic of 1000+ requests per second probably doesn’t account for much of the budget, not to mention the platform can’t even deal with a traffic of 100 requests per second. Anyway, I am not an expert in DevOps and I don’t know how much actual cost was spent on cloud fees such as servers. Maybe the government can provide actual breakdown cost items and relevant data to the public so technology experts or experienced engineers can judge whether the cloud cost is reasonable and if is there anything wrong with the platform’s DevOps management.

Although Bill Gates only talks about one U.S state health department’s need for upgrading software, the U.S also encountered the same problem. In the U.S, the federal government let state governments come up with their ways of distributing vaccines. Washington state developed a website, and there was a crash when over 36,000 people were online ^[28]https://www.cna.com.tw/news/firstnews/202103170290.aspx . Again, at first, I thought it was a problem that only happens in Taiwan, but now I realize that the U.S has the same problem when their big tech companies such as Google can do far better jobs. Back to the topic, although I don’t know the actual timeframe for the 36,000 people, technology expert Raphael Lee stated, “No website of this scale should fall over when 37,000 users are trying to access it.” He said, “Those numbers are not that big. ^[29]https://statescoop.com/washington-dc-vaccination-website-crashes/ ” What’s more, according to Wall Street Journal, the reason why there were low-quality platforms in the U.S is that the state governments often are under-resourced when the federal government lets states come up with their ways of distributing vaccines. “States are under-resourced, and don’t necessarily have the resources to build websites to support a vaccine rollout,” said John Brownstein, an expert ^[30]https://www.wsj.com/articles/health-officials-scramble-to-provide-booking-systems-for-covid-19-vaccines-11610488516 . What I want to let you know is that at least Americans admit that the platform sucks and there were technology experts who spoke up it indeed sucks, and the root cause of the problem in the U.S is because local governments are under-resourced. On the contrary, many Taiwanese aren’t willing to admit our platform sucks; there is no technology expert to show up and speak out, and the Taiwan government had enough money and resources to make a decent platform, but it didn’t.

Back to one of the doubts from some Taiwanese, some of you think the legislator Mr.Zheng isn’t a technology expert, so she shouldn’t criticize the vaccination platform? I wonder why only technology experts or engineers can speak up because the platform indeed sucks that even a non-technical person should be able to perceive. Although I speak a bit more technically on this topic, I am not an engineer, either. There have been many people across the world requiring an expert to criticize a specific topic, otherwise, they think “Who the hell do you think you are?” In my view, there is no need that it must be a relevant expert to criticize a specific topic; there is no need that it must be a technology expert or engineer to criticize a platform. They may not able to provide appropriate suggestions on how to improve the issue they criticize, but they still can criticize a platform or a specific topic – it’s just that it’s difficult for normal people to judge whether the criticism is reasonable or not, and most of the time, it’s experts who find out problems and provide criticisms. This is a point that I hope everyone who reads this article learns. In addition, some U.S technology experts spoke up after the crush events of vaccination platforms, but there were no such experts who spoke up in Taiwan, including Taiwanese engineers who work in Google’s subsidiary in Taiwan. I hope when any criticism of technology topics occurs in the future, technology experts and experienced engineers in Taiwan can proactively and publicly speak up and provide their perspectives. Even if you are not a famous engineer, you can just say, “I am an engineer, and I can achieve this requirement if there are [how many] persons who also can achieve this requirement work with me and sufficient budget for cloud fee”. Taiwan media have to interview some technology experts and ask for their perspectives, too. Although there is a company, 91APP – a Taiwan eCommerce service company that is the only Taiwan eCommerce provider that withstood huge traffic during the Double 11 festival ^[31]https://buzzorange.com/techorange/2018/11/15/how-does-91app-deal-with-1111/  – can deal with 8 million tweets in 8 minutes according to an article from the CPO of 91APP – that’s 16,667 requests per second, which is great ^[32]https://medium.com/retailscience/2020雙11觀察報告-46625a79e049 , another major eCommerce company in Taiwan, Pchome, probably can’t even deal with a much smaller amount of traffic. The average traffic of Pchome is 32.44 million per month ^[33]https://www.top10.com.tw/life/938/top-10-online-shopping-website/ , which is about 12.5 requests per second, and the traffic during the Double 11 festival is 10 times of that of normal days, according to this news which states that because the traffic of Pchome increased by 10 times, the system went abnormal ^[34]https://udn.com/news/story/7241/5884010 . Assume the average traffic is the traffic on normal days, by calculation, Pchome dealt with traffic of 125 requests per second in the Double 11 festival. If a major company in Taiwan can’t even deal with a traffic of 125 requests per second, maybe most of the Taiwanese engineer’s ability can’t deal with such traffic, too. If that’s the case, no wonder no one jumped up and said “I can accomplish this requirement of traffic if there are [how many] persons who also can achieve this requirement work with me and sufficient budget for cloud fee”. Still, Taiwan media have to interview technology experts and ask for their perspectives, so we can know the real ability of average Taiwan engineers, you know.

There is another concern from a Taiwanese: It’s possible that the APIs from the National Health and Welfare Bureau for validation may not be able to withstand such high traffic. If this is true, yes: the responsibility for this crash problem will not belong to the development vendor. However, the problem will become that of the engineers who developed the APIs from the National Health and Welfare bureau and the government.

7.(3).2 Integration with hospital systems

A Taiwanese pointed out that it took a long time for developing a vaccination website. It should already take a lot of time in integrating with systems of local health bureaus, hospitals, and clinics. Nevertheless, according to an article from Common Wealth, the vaccination website had not been integrated with hospital systems when there was information that the vaccination website is a one-stop service. Many citizens who finished the appointment of vaccination searched hospital appointment systems but found out there were no appointment data in hospital systems. The Taiwanese asked, “So what has the development vendor been busy with all these times?^[35]https://www.dcard.tw/f/2019_ncov/p/236475248 ” Another Taiwanese rebutted the same complaint on another website by stating, “Every hospital system is different, how can the government integrate with all of them?”^[36]https://www.ptt.cc/bbs/Gossiping/M.1626153064.A.9FB.html 

That’s a good question. Integration with all hospital systems in Taiwan is a huge task. According to the Ministry of Health and Welfare’s data, there are about 17 local health bureaus and 479 hospitals in Taiwan – that’s 496 systems, it’s indeed a huge task ^[37] … Continue reading. Nevertheless,the development vendor of this platform doesn’t need to integrate 496 times with all of the systems; the vendor only needs to develop APIs which provide appointment data for engineers who work for the hospital and therefore is responsible for maintaining the hospital system; it’s the job of the engineers who work for each hospital to call the APIs developed by the vendor to get appointment data and let data be automatically and real-time recorded in their hospital systems. Therefore, the vendor doesn’t need to do integration as long as it develops and publishes its APIs of the platform, and engineers of each hospital only need to integrate once for integrating only the system they are responsible for. (By the way, I was almost fooled by this concern of so many systems as I have stayed away from software topics for a long time.)

The government also needs to discuss official API columns which are acceptable by the majority of the hospitals with the hospitals. Some of you will ask, some hospital wants these data, while some hospital wants the other types of data…it’s very difficult to deal with. This is probably a question that will come from many hospital managements and Taiwanese engineers. So when we encounter such a problem, we just give up?

No. In my view, that’s not a problem. The API columns don’t have to be suitable for all hospitals; being suitable for the majority of the hospitals is enough – the remaining part is each hospital’s business. Why? Because as long as a hospital has the patient’s unique number of the health insurance card, doctors are able to get the patient’s relevant data including medication records with that card number, which is what they need. Names, date of birth, the unique number of the health insurance card, and the phone number – it’s right that these are mandatory data that need to provide to hospitals, but for some data which some hospitals need to get while other hospitals don’t, it’s the problem of the hospitals which needs those data. Deciding what types of data are needed is actually simple: Just provide all types of data that the health insurance bureaus need to check when they do audits and additionally add the phone number as one of the data needed – these data definitely already include Names, date of birth, and a unique number of the health insurance card. The only data which needs to add on is the phone number which is not listed in Article 12 of the Physicians Act(so I am not sure if a phone number is mandatory, but it should be mandatory), so hospitals or clinics can contact patients when they need it.

The integration work by the hospitals will still be difficult. Why? Because many hospitals are still using outdated websites for patients to make appointments, judging from the current state of many outdated hospital websites, the reason that they are outdated is either a lack of up-to-date skills from engineers who are currently responsible for these systems, or a lack of funding for hospitals to do updates for their websites, or both. Therefore, if the problem is a lack of up-to-date skills from engineers in the hospitals, such engineers probably aren’t capable of doing integration jobs, either. Then the integration jobs will probably need to fall on the heads of the development vendor of the vaccination platform, which may cause some headaches. There is another situation where some hospitals outsourced the development of websites to software vendors. It will be a problem if these hospitals lack of funds to ask their vendors to do the integration job. Nevertheless, it’s still worth developing APIs for saving a lot of human resources for hospitals in getting data or saving efforts from patients who register their data, and developing APIs is still the best solution at the current time to integrate with various systems.

It’s important to note that it seems that the development vendor didn’t think of developing APIs – probably because the developers are inexperienced. The vaccination platform seems to have done the integration after then (simply the integration job should be finished by the time the platform was rolled out but it wasn’t.) Nevertheless, according to Common Health, when the hospital called the Health Bureau to ask about the one-stop-appointment platform, all the health bureaus were confused. They said, “If an appointment job is done in Ms.Tang’s vaccination platform, does this mean that the appointment job is done successfully? Or does a citizen still need to make an appointment on the hospital’s website again? No one can figure it out.^[38]https://www.commonhealth.com.tw/article/84651 .” Judging from the words, it appears that hospitals and local Health bureaus didn’t know anything about the integration work. Then the integration job is probably not done by developing APIs and letting engineers in the hospitals and local Health Bureaus do the integration jobs, because if that’s the case, hospitals and bureaus shouldn’t have the confusion of “Should a citizen still need to make an appointment in the hospital’s website again? No one can figure it out” – this sentence looks like they all thought the integration jobs should be done by Ms.Tang’s vendor. If the integration job was indeed not done by developing APIs, I will want to ask why even Ms.Tang, a software expert, didn’t think of developing APIs as a solution for integration with 496 systems. Using APIs for integration is a common practice in the software industry, which Ms.Tang as a “software” expert should know. Many software engineers have used Google or any big tech company’s APIs to implement in either their work projects or side projects – you probably used one of those APIs, too. If using Google’s APIs seems like an unfamiliar topic to you, there is another example: Many Taiwanese engineers use LINE’s APIs to develop robots as their side projects; many of you should be familiar with this example. Anyway, the vaccination platform and many counties’ platforms have integrated with hospital platforms after then. I don’t know how the developers did it; I simply guess based on the words in the news. The above paragraphs simply have two purposes: One, answering the question “Every hospital system is different, how can the government integrate with all of them?” Two, since integration with 496 systems is possible by developing APIs, the Taiwanese’s complaint and the question of “Why it took a long time to develop a platform but the integration job hasn’t been done yet” at the time when the integration job indeed hadn’t been done yet is reasonable.

In addition to the development of APIs as a solution, two points should be noted. First, even if we assume that there is no such an API thing in the world and the development vendor has to do integration 496 times on its own, didn’t the vendor agree to do so? Even if the vendor’s estimated time of completion is too optimistic, shouldn’t it at least complete integrations with local health bureaus or a few hospital systems? But it didn’t. What’s more, the Ministry of Health and Welfare didn’t even tell the public that the integration with hospital systems isn’t completed, either. If there is no development of APIs as part of the plan, the Ministry of Health and Welfare should tell the public that the integration with hospital systems hasn’t been completed because it agrees to do integration. In the future, if the development of APIs becomes part of the plan but the integration job by some hospitals isn’t successful, the Ministry of Health and Welfare should also just tell the public about how things are going. Second, the system without integration with hospital systems is far from what the ideal platform is supposed to be. How to achieve the ideal state? Other than the abilities of engineers, conditions for achieving the ideal state are about whether the budget can satisfy the requirement of integration with 496 systems if there is no such an API thing in the world, and whether the limited time is enough for completing such work or not. In my view, even if there is no such an API thing in the world and the development vendor has to do integration 496 times on its own, a 200 million budget should include jobs of integrations with some hospital systems, but I am not sure if it’s enough for integrating 496 systems (That said, it should be possible – simply I am not sure). Yet, lack of budget shouldn’t be an excuse, either. The development vendor or the Ministry of Health and Welfare should either raise the budget for integration with 496 systems or reduce the budget (make it less than 200 million) for removing the requirement of integration with other platforms. Without integration with hospital platforms, 200 million is too expensive, therefore it is still wasted. Anyway, there is such an API thing as a solution for integration, and it’s important to note that 200 million is enough for including the development of APIs.

7.(3).3 Interface

Another problem with the vaccination platform is its interface (The view you see from a screen).

7.(3).3.1 The interface is not user-friendly.

A Taiwanese pointed out, “Since the registration of the preference on types of vaccines is done, why do we have to go to the system again to make an appointment for inoculation? Why don’t they just automate all of the things? After registering the preference on types of vaccines, the platform makes the user in the queue of inoculation. When the vaccines arrive, the platform automatically notifies the user of the time and date of the appointment. Isn’t this way more convenient? ^[39]https://www.dcard.tw/f/2019_ncov/p/236475248 ” This suggestion makes sense and I will only want to add one thing: the registration job should also include a user’s available dates and timeslots, then the arranged time of an appointment by the system will be more convenient for users. Nevertheless, let’s also see why the government wants to have such a flow. According to Common Wealth Magazine, there are two reasons that the government adopts the two-step flow of registering first and making an appointment later. Point One, the arrival time of different types of vaccines varies. Opening preference registration is good for the government to control the preference of civilians. When the stocks of a specific type of vaccine are enough for covering the number of registrants, the system can start to notify registrants to make an appointment. Point Two, the order of registration doesn’t impact the order of inoculation. If the registered time is used as the basis, people who know how to use digital tools may take the lead. Therefore, in the future, when the government notifies the reservation, it will follow the principle of “respect for elders” and use the year of birth to arrange the order of inoculation ^[40]https://www.cw.com.tw/article/5117434 .

The concern of avoiding the older people who don’t know how to use digital tools fall behind makes sense. Washington state encountered a similar issue and changed its flow to one that is close to Taiwan’s. Because too many people rushed to book an inoculation so it was difficult to book an inoculation, in 2021/3/10, the Washington state government started to change into a pre-registration flow. Appointments will be made on a case-by-case basis according to the amount of vaccine assigned each week, the priority of people receiving the vaccine, their area of residence, and the nature of their work ^[41]https://www.cna.com.tw/news/firstnews/202103170290.aspx .

Yet, from my perspective, a vaccine platform still shouldn’t have such a two-step flow.

First, concerning Point One that the government wants to wait until the stocks are enough for covering the number of registrants,

NO, the proposed flow is still possible for the government to wait until the stocks of a specific type of vaccine are enough for covering the number of registrants because the platform makes the user in the queue of inoculation. Notice that the platform should automatically notify a user what an arranged time of an appointment is: Just make the platform waits and automatically arranged appointments when the stocks are enough to cover the number of registrants. The arrival time of vaccines is unpredictable. How can a platform know when vaccines will arrive? Make functions in the back office of the platform that allows government personnel to be able to record “[a specific number] of [a specific type] of vaccines are arrived on [Date], and they are distributed to [XXX] county”.

Second, concerning Point Two that the government wants the order of inoculation to be based on “Date of birth”, not “the order of registration of preference” as it is afraid that people who know how to use digital tools may take the lead, NO, this solution has a serious problem. After older people are prioritized and notified to make appointments, isn’t an actual appointment time still based on the order of making the appointment among old people? It’s just that it’s about the order of doing the second step. Yes, older people are indeed prioritized, and I agree that they need to be prioritized, but there is a better solution. That is making old people prioritized in the first step of registering preference. Adding age as a prioritized criterion in the first step can easily solve the problem. You see, either way, has to let old people use technology, or they just go to pharmacies for making an appointment if they don’t know how to make an appointment online. So why not make the flow simpler by adding age as a prioritized criterion and removing the second step? In my opinion, since the person who design this system and people who agree with this complicated flow didn’t think of adding criteria in the first step, they have serious problems in categorizing and arranging things, and in logical thinking.

Washington’s case is the same. Just make functions in the back office for government personnel to record and distribute the available number of vaccines, add criteria such as the nature of their work and age, and make people in certain occupations or a certain age prioritized in the first step, is sufficient to arrange inoculations while preventing chaos.

7.(3).3.2 Interface aesthetics

The interface of the vaccination platform in Taiwan is not beautiful, compared to the modern website or Apps we often use. The U.S platform from CVS Pharmacy did a much better job. You can see some of the interface screenshots from this article. In fact, the interfaces of the government’s many digital tools were ugly. This is the interface of the Health and Welfare app from the government in 2020 to book masks (Click the link). It was until 2021 that the government made an update so now we finally have an APP with a beautiful interface. I am just complaining that why we have an ugly interface with that APP in 2020. Many interfaces of APPs we often use, for example, Facebook, or banks in Taiwan already have very modern and beautiful interfaces before 2020. When I saw the interface of the Health and Welfare app from the government, I thought I was living 20 years ago, a time when most websites don’t focus on aesthetics of the interfaces. The interface aesthetics definitely needs to be taken into account when the government plans to develop any other website or App in the future.

Although I said the U.S’s platform did a much better job, that’s just one example. Some platforms in the U.S are actually ugly, too. The interface of New York’s Vaccine Finder is one of the ugly-interface examples ^[42]https://vaccinefinder.nyc.gov . The reason that the interfaces of these U.S platforms seem outdated is probably what Bill Gates said in his book: Lack of funding and resources of state governments. Nevertheless, the platforms in Taiwan which I pointed out, come from the central government, not the local government, and the central government indeed have enough money to make a proper website or APP (200 million). So the Taiwan government has a serious problem in reviewing these digital tools when reviewing them even doesn’t require a certain level of interface aesthetics. I suggest the Taiwan government take a look at the interfaces of other beautiful APPs.

Read Part 6

If you like this article, please share the article to your social media page, so my article can be accessed to more people. 
Please also follow me on social media by clicking the links at below, so my latest articles can be reached out to you.
Follow My Social Media: Facebook | Twitter | Linkedin

Reproduction of the article without permission is prohibited.

Chrissy Slora

The articles published in this website are all written by Chrissy Slora. Chrissy Slora aims to make people learn about how the world can be better and provide her intelligent and wise insight by writing about various topics from technology to issues happening in the world. Follow her on social media: Facebook | Twitter | Linkedin