So what’s with the NHS track and trace app? Initially the NHS decided to build their own Bluetooth mapping initially rejecting the Apple and Google track and trace [link]. Their bluetooth tech could only detect 4% of iPhones, and 75% of Android devices [link]. This cost £10 million [link].
For those who are not in the investment side of tech, this is expensive. For example, our startup had £2 million seed money initially. For £10 million you can buy some really, really good tech. A trial that fails to detect a nearby iPhone 96% of the time after spending £10 million is not a good sign.
After this, the NHS swallowed its pride and used the tech that Apple and Google had developed. They still spent over £35 million [link], however, the app seemed at face value to be pretty good. It’s decentralised, the codes are regularly refreshed, and the algorithm is designed to be fault tolerant. It only alerts people who are high risk. This is people who have been within 2 meters of a positive person for more than a certain amount of time. However, how it is getting on now?
First of all we have to consider compatibility. We try and do as much work on the servers as possible because writing code for user devices is a headache. When I was working in financial tech, one of the pains was ensuring that our code ran on old versions of internet explorer, because customer service desks ran this and were not allowed to update it. As soon as I hear of an app using hardware like bluetooth, I instantly cringe about old phones and operating systems not being able to support it. Considering this, it’s not a surprise that a lot of over 60s (generally some of the most vulnerable) cannot download the app because their phones are too old [link].
This leads onto the next issue, the network effect. This app’s value is derived by network size. If only one person downloads the app, it’s completely worthless. The university of Oxford concluded that 60% of the population needed to use it and this was widely reported. However, it isn’t an all or nothing and a lot of reporters oversimplified this statistic. The Oxford simulation [link] showed that we only need 14% of the population downloading the app to see positive effects:
With a total population of 66.65 million and 79% of the population having access to a smartphone [link], we will need 17% percent of smartphone users to download the app. Considering that 10 million downloaded it [link] that’s 18% of smartphone users. Ok it’s not a roaring success, but it managed to get 1% above the minimum threshold. It doesn’t say how many uninstalled it etc, as I’m sure a few will have downloaded it out of curiosity only to block it accessing parts of the phone or uninstall it later on. Considering this, if I was PR for the app, I’d be working on how to show it did some good and it wasn’t that expensive as opposed to trying to convince the public it was a great success. This is poor even if the conditions are perfect, however, they are not.
Bluetooth still adheres to the laws of physics and different scenarios such as rooms and building materials give out differing results. Also not all hardware in the phones are the same. resulting in a ton of false positives [link]. Bluetooth works along the same frequency band as microwave ovens. Out in the wild, there’s a whole host of interferences. Concrete walls vs glass windows have a different effect of how the signal is traversed. Bluetooth is good for detecting devices. However, considering all the variances, it’s a completely different ballgame when detecting distance.
So, is all this worth it? Probably not. Straight off the bat with the 10 million downloads at a 35 million cost is 3.5 million per user. With this you have to consider an economics concept called discounting. The basic premise is that time is money. £2000 now is more valuable than £2000 in two months time. That £2000 could have been put to work in those two months buying machinery to carry out tasks etc. This is why finance options for products are not always a rip off. If what you’re buying is going to produce or save labour over a period of time that is worth more than the interest on the finance, then it’s worth it. Building an app takes time, and testing. The longer it takes, the more of a waste of money it is, simply because you could have put that money to work in that timeframe. With pandemics time is also important. However, we shouldn’t be too damming at the cost. Yes government is famous for being inefficient, however, we have to note the scale. There’s 223 NHS trusts. If we split that 35 million to those trusts, that would only result in each trust getting roughly £157,000.
I get the desire for the app. The economy has been damaged via lockdowns, and a dynamic tracking system could have enabled us to get the economy up and running, surgically removing parts for two weeks where hotspots occur could stem the pandemic whilst avoiding another lockdown. However, finding data on the success is hard due to the fact that the app is decentralised as opposed to centralised. This decentralised approach was chosen to protect peoples’ data. Considering the blanket lockdowns of whole areas I’m guessing the app isn’t having a huge effect. There have also been multiple places of work and schools telling their attendees to turn the app off when in their institution [link][link][link]. Whilst it’s easy to be outraged, we have to put ourselves in their shoes. The app is known to shoot off false positives could potentially shut down large chunks of your business for a couple of weeks. I would love to pretend to be a selfless hero but in this economy, I’d be thinking along the same lines.
I myself personally haven’t installed it and I’m not going to. I don’t know any of my tech literate, or medic friends who have. Some of my friends in key institutions like Oxford and Imperial who are working on COVID projects have told me they’re not downloading it. Wearing a mask, washing hands, and meeting friends in non crowded areas has resulted in me not even getting a cold. There was one scare where a friend had cold symptoms a couple of days after meeting. I messaged other friends who I met after that, and we stayed in doors for two days until the test for the coughing friend came back negative. The king pin here is adhering to hygiene protocols.
So what could have been done? I think a lot of people fetishise tech. I love tech…. I’d be in the wrong field if I didn’t. However, it’s as useful as the people using it, and it does not defy the laws of economics. Also, developing software is hard. You have to consider security, data management, edge cases in the code, user factors, scalability etc. We are spoilt with great engineering from companies like Google, Amazon, Microsoft, and Apple. However, these companies have been building their products over 10s of years, and hire world class engineers. For instance, a senior software engineer in London will be pulling a range of 80k-150k with stock options and bonuses [link] after 3-5 years experience. If it was easy, and anyone could do it, greedy companies would not pay those levels. Big household names go for even more. Uber signed the software engineer Anthony Levandowski for $680 million [link]. He was later involved in a scandal of trade secrets with Google but the price tag is not to be ignored. The point here is that we are spoilt with good engineering. We turn on a phone, hop on the Facebook app, check out Gmail, and then order something on Amazon before switching on Netflix and binge watching a boxset only to order something off Deliveroo when we’re hungry. To a casual user they just work. However, under the hood it’s not straightforward. Google has a lower acceptance rate than Harvard with entry-level engineer pay over $150,000 a year [link]. The selection process is brutal with multiple interviews, whiteboard tests (producing an algorithm to a problem on a whiteboard in-front of interviewers), exams, and coding challenges. Without this appreciation, many politicians fall into the trap of thinking that software is easy, and that we can knock up a solution in no time. This isn’t helped by conmen promising the world.
So what do we take from all this? Whilst we shouldn’t be jumping up and down proclaiming that the money spent of the app could have completely changed the course of the pandemic, I think it could have been better spent on early intervention of PPE or COVID compliance officers. The idea of a track and trace app is a good one, but knocking one up in under a year is unlikely to be useful. We’ve now learnt the need for track and trace technology. A long term global development of the tech over a couple of years in a modular fashion is the best approach. An example to draw from is blockchain. Years are going by and central repositories for the ledger technology are growing. Amazon web services has now under beta started offering blockchain in certain regions of the world as a bolt-on service. With blockchain, we’ve seen problems with seed generation, scaling, privacy etc. They are slowly getting ironed out as the years go on. Decentralised track and trace will ideally take the same trajectory.
The idea that track and trace would work in this pandemic is the result of naive politicians and people believing their own hype (remember when Elon Musk offering a submarine to the rescue divers). We have to come to terms with the fact that we end up in positions where there isn’t a silver bullet. Hopefully this pandemic has taught us that cramped living conditions of the poor [link], access to technology for vulnerable people, and a united WHO preparation and response to outbreaks globally are in everyone’s interest.