The contact-tracing app, accuracy?

Go on, i’m curious…. how can they manipulate society with this data?

It’s like the scene in the action movie where the prison’s exploding because the bad guys are attacking and the mass murderer (who’s perhaps the good guy now) asks the guard for his gun to fend them off.

Does he:
a) give the gun to him to fend them off,
b) give the gun to him to fend them off but is then shot, because he’s still the bad guy
c) give the gun to Chuck Norris who’s just out of shot shouting “don’t give the gun to him I’m a better option, roundhouse kick to the face”

Seriously, how the **** does having location data (which already exists) on someone influence an election.

TINAS beat me to it. Get enough data and the assumptions you can make from it are remarkably solid. Certainly enough to manipulate (or target if you like) small and otherwise uncoordinated groups of people (who may only need a wee push) to think/vote the way you want en masse, by targeted (or hidden from scrutiny) ads.

Aiui the only time the app logs data to the central server is if you show symptoms. So most people are unlikely to ever export any data (remember the idea here is to stop the spread) and those that do will likely only report once or twice. If you wanted a system to gather mass data on a population then this is a really bad way of doing it.

Nope, that would be the decentralised system, based around Apple-Google API, not NHS centralised one.

A-G only the anonymous keys associated with a positive test get uploaded, and all phones check to see daily if there are any new anonymous keys to check locally on the phone for any potential match. And that potential positive is only a warning to the phone holder.

With NHS system, afaict, all “anonymous” keys and all contacts are uploaded for the server to crunch who may be potentially infected.

No way in hell is that app going on my phone, but then fortunately I’m in Germany where they decided that the non-centralised version is more likely to get uptake than one that knows everything.

What are ‘they’ looking to steal/do ?

As another quick little example, relevant to this forum.

This in isolation has little utility:

However, it can then be used to create this:

https://www.strava.com/heatmap#7.00/-120.90000/38.36000/hot/all

OK, so we have the ‘big data’ influencing politics thing. And the 24/7 state surveillance worries.

IMO what they actually want to do is have a nice UK map back at ukgov/PHE/NHS mission-control that displays where people are reporting symptoms and getting positive test results, i.e. a near real-time feed from the contact tracing app of where outbreaks may be happening so they can deploy resource. Very much dependant on decent takeup of the app, quick-result Covid tests in the post, the tech working etc.

You can’t do that with the completely distributed model. Note they’d still be able to do that using Apple/Google, people have pointed out you can do what you like in your app that sits on top of the Apple/Google API (except using GPS).

IMO what they actually want to do is have a nice UK map back at ukgov/PHE/NHS mission-control that displays where people are reporting symptoms and getting positive test results

The positive test results can be handled at the test centre surely? As soon as test returns positive, enter it into the system. No need for an app and it’s verified and instant?

The positive test results can be handled at the test centre surely? As soon as test returns positive, enter it into the system. No need for an app and it’s verified and instant?

Yes, but when you’re working at this scale (or the scale we will need to be working at) any automation is good. And tests may not need labs. The US keeps mentioning their self-test that gets a result in 15 mins for example.

A bit like the (is it an urban myth, IDK) story that Boots rewards card system can tell

That will have come from a story about Target in the US.
Not sure the story has been completely checked out or if it was one which has just been repeated so often its claimed to be true but essentially a man walks into the store complaining that they sent some baby clothes vouchers to his teenage daughter. Obviously they apologised but when a senior management called up a few days later to follow up the bloke apologised to them.
Apparently it was stuff like buying lots of unscented lotion and a few other things strongly correlated with being heavily pregnant. So if that suddenly changed it was advert time.

Your likes, interest and how you think etc I understand but locational data? Think theres a bit of over thinking it going on here.

TINAS got it.

Imagine if you knew that people who drink at a pro-leave pub chain like Wetherspoons were more likely to vote for Boris. Then all you have to do is find out where their customers mainly live and then target that area with leaflets aimed at their ideals and morals. The areas where few people drink at Wetherspoons will fit some other demographic, say customers who shop at Morrisons, were more likely to be passionate about elderly care in the community. You can then send leaflets about that to them. Then another area where people who go to the local art house/theatre will respond to info about how you support the arts, send them leaflets that show you’re good at that.

Use all of that over multiple areas and you have different areas and demographics voting for you for different reasons. You can pretty much tailor your national campaign to suit every local issue without alienating any others. For a working example see the Vote Leave campaign from a few years ago. Wonder who was behind that?

For a working example see the Vote Leave campaign from a few years ago. Wonder who was behind that?

Well exactly.

It is so easy to sit here and marvel at how so many people in the US cannot see that Trump is a disgrace an embarrassment and a real danger.

But a lot of those folks aren’t the most adept at ‘critical thinking’ (obviously I could be unkind and remove the word ‘critical’ but I won’t).

They do not seek out a second or third opinion and they do not sit down and watch the news at an appointed time like ‘the good old days’ – even if the US news scene is an absolute joke.

They are fed their own prejudices and hatreds back to them as ‘news’ via targeted media.

And we are nearly as bad here.

Seriously, how the **** does having location data (which already exists) on someone influence an election.

Your likes, interest and how you think etc I understand but locational data? Think theres a bit of over thinking it going on here.

As above what location data linked to a postcode has some very powerful applications. Politically, you will know which are the most sensitive seats. All parties will know that. With this info, you could easily (well not easily, but it’s achievable) build a model to show how to maximise your advertising capability to your specific targets. Then, as above, you could maximise the impact of each one based on more localised trends. Opposition parties will not have this info so will be far less efficient.

TLDR; Politics is about getting the right message to the right people. The more you know about people, the easier that is.

I don’t buy the Cummings evil puppet master stuff, but I do have concerns about the way that the app is centralised and collects data.

I don’t have anything to hide, but my privacy is important to me. That’s why I make the choices that I do about the platforms and technologies I choose to use and not use.

The case is for the people developing and maintaining and overseeing the app to explain and reassure me about how the data I will provide be used, stored, and if ever deleted.

When there seems to be a solution provided, which a. Is more private and b. Probably more effective, the above communications need to be even more clear.

Once you give up your privacy it is hard to reclaim. If they want this to work, the government need to address these concerns.

It is so easy to sit here and marvel at how so many people in the US cannot see that Trump is a disgrace an embarrassment and a real danger.

But a lot of those folks aren’t the most adept at ‘critical thinking’ (obviously I could be unkind and remove the word ‘critical’ but I won’t).

Trump is, I think, the most incredible politician of our lifetime. Obviously a monster, but even so. And the mistake I think you make is believing it’s only those short of critical thinking that are being manipulated.

and if ever deleted.

It won’t be. You won’t even be able to ask that it’s deleted.

IMO what they actually want to do is have a nice UK map back at ukgov/PHE/NHS mission-control that displays where people are reporting symptoms and getting positive test results, i.e. a near real-time feed from the contact tracing app of where outbreaks may be happening so they can deploy resource. Very much dependant on decent takeup of the app, quick-result Covid tests in the post, the tech working etc.

Well yes, but as soon as you tell the app you ave symptoms it will need your address to send the home testing kit to. Or you’ll go to a testing center and fill out a form.

Then the App can do it’s job and ping those people you’ve been in contact with.

I’m guessing it won’t tell people to self isolate just on the basis of someone they shared a buss with reporting a cough, it’ll need to be a bit more concrete than that. Also, unless you ping all those people on the bus and get them tested you don’t find out where the outbreak is anyway because you don’t see all the asymptomatic people.

FWIW I’ve got relatives in a small town of ~14,000 in Canada, they had an outbreak, and within days had split the population into 3, those that were symptomatic, those that were likely exposed via contact tracing, and the low risk, and tested everyone in that order. So within days you know whether you need to be staying home or not and the town goes back to normal-ish because the outbreak is contained. That’s how it should work.

Just like being on holiday in the US and using 90210 as a zipcode, we could presumably use a different postcode to sign up, just to keep the bastards on their toes…

I don’t have anything to hide

It’s absolutely not about this. I think this is most definitely where the Tories will try and probably succeed to convince folk to use the NHS app, they’ll say exactly this, and they’ll be correct..But hugely disingenuous.

It’s about putting folk into demographic groups to target information at the group level. Not even Cummings can target ads at an individual level (although Lord knows he’d love it if he could) . Your personal data isn’t really at risk. But OUR data collectively; absolutely is.

Ironically it’s a lot like the virus; the question isn’t whether you as an individual are at direct risk; it’s about the big picture and how everyone affects everyone else

Did a bit of digging, the location priv is necessary for Bluetooth LE. They should just be more transparent about it

I did wonder if it was that, the advantage the iOS/Android api has is it’s not going to panic people with the location permission as it’s a Covid permission.

I just think any app that triggers location permission messages in any permission requirement on installation isn’t sending the right message for a government app.

It’s all about trust for me.

Suggestion here that they are now considering switching to the Apple/Google design.

FT exclusive: NHSX has asked its software developers to "investigate" switching its centralised contact tracing app to Apple and Google's decentralised system, according to contracts seen by @FT. w/@SarahNev @helenwarrell
https://t.co/5G7XLskp4G

— Tim Bradshaw (@tim) May 6, 2020

Turns out the existing app might actually work:

https://mobile.twitter.com/jasonkneen/status/1257790047715315718

And:
https://reincubate.com/blog/staying-alive-covid-19-background-tracing/

(Testing conducted on iOS)
So if they’ve got it going on Apple I imagine it does with Android.
Still, if they adopt the Apple/Google path after all I’ll be much happier.

Good evening,

Thank you for continuing to tell us about your health daily. We are now 3 million strong and have some amazing accomplishments. As we start to come out of lockdown your continued support is going to be more critical than ever.

You’ve helped us to predict COVID hotspots, demonstrate to the government that lockdown was working, and identified loss of smell as a key new symptom of COVID. This week, we have agreed to support some important NHS research by adding a few new questions to the app.

An important update
Before we tell you about some of the new questions we’re asking, I wanted to let you know some news about our name.

We are changing our name to the COVID Symptom Study. We do not follow you around and we felt the use of the word “Tracker” was causing confusion with apps that track and trace. We are a community of people working together to help save lives and support front-line NHS workers across the country. Through your continued daily reporting you are making a truly important contribution to understanding this terrible disease.

With your help, over 25 scientific papers are already being published which will help doctors and scientists globally to beat the disease. I hope that like me you are proud to be part of this. To discover more about what you are making happen, do visit the blog. We will update this each time we publish a new paper, and explain what has been discovered.

New Questions
We recently added an ethnicity question as you may have seen, which is aiming to try and answer the very important question of why people from ethnic minorities seem to be much worse hit by COVID. By understanding the science, we can enable action.

Tomorrow we will add some new questions. We have been asked by NHS England if we can help them to understand why men are more at risk from COVID than women. We have therefore teamed up with NHS researchers Dr. Louise Newson and Professor Janice Rymer, who think that oestrogen may play a protective role in fighting COVID. To understand this better, we have added a few new questions in the app that we can then link to the unique symptoms that 3 million of you have been experiencing. These questions look at things like contraception and Hormone Replacement Therapies (HRT) that impact hormone levels.

We expect to add further questions in the future, as we continue to focus on how together, we can help save lives.

Tune into our webinar tomorrow at 5pm BST on YouTube to hear Dr. Louise Newson and Professor Tim Spector explain this groundbreaking research.

I look forward to sharing more news with you shortly. We have some very exciting things planned for next week, as we focus on how together we can get out of lockdown safely and hopefully faster.

Thank you again for your support.

Professor Tim Spector

Hopefully this helps persuade some people that this isn’t about politics.

It’s not the NHS app but the guys from here https://covid.joinzoe.com/

It could be something random like people who ride bikes but also go to Burger King are more likely by 0.3% to Vote one way than people who don’t ride bikes but got to BK. But people who ride bikes but also go to Primark are 0.05% more likely to vote the other way. Add in enough of those correlations and you can start to make some remarkably solid predictions.

So back to my point, that is all around interests and the mindset of individuals of what they say/like etc on platforms like Facebook. They use that data as you’ve pointed out to determine demographics. Those methods are then applied on platforms such as facebook, instragram etc. Which disregards regional borders, it’s advertise to any user to who views that platform.

That we are in agreement with.

I am still failing to see how they knowing I was in Tescos for 50 mins or the 2nd wifes for 3hrs yesterday adds to what they already have and can help further manipulate my mind.

They know I’m based in Kirklees, they know I work in Halifax, they know I was born in another part of the country and so on, what macro level of locational data for a small subset of time based on a period whereby rules are in place to restrict movement to a degree and therefore influences my decision to go to places, be of any use in the election in 4 years time?

Regardless it doesnt record GPS data anyway –

The NHS app uses Bluetooth signals to check for contacts. This means it does not need to connect to mobile data when you are out and about. For now the app does not track GPS signals.

The NHS said it might introduce a system of monitoring location data in future to collect useful data on the pandemic, but that would be voluntary.

Like so many things that our government does these days, the app is a red herring as far as your data is concerned. A talking point so we are distracted arguing amongst ourselves.

What you need to be worried about is what they are doing with existing data under the guise of app development.

https://www.buzzfeed.com/alexspence/boris-johnson-dominic-cummings-voter-data

Exclusive: government blocks full publication of AI review

https://www.computerweekly.com/news/252482365/Privacy-International-puts-Palantir-in-the-dock-for-NHS-data-analysis-work

Etc.

Data is power, if it wasn’t, no one would bother collecting it. If you have access to data and data sets about customers that your competitors don’t have access to, then you have the advantage (provided you know what to do with it).

I am still failing to see how they knowing I was in Tescos for 50 mins or the 2nd wifes for 3hrs yesterday adds to what they already have and can help further manipulate my mind.

From that one piece of data; **** all I’d imagine, but that’s not what data manipulation is about. To use the “going to Tesco” analogy, they’re looking at How many times you go there? How long is each visit? Are you going for the weekly shop? Men don’t normally do that, so what groups of men do? and do you fit that profile? and what can we infer about you that matches what we know about men who shop for x times a week, for y mins at a time?

If there is a profile, can we direct a message that we know from testing that resonates with Men who fit this category.

It message might not work for you, but it will work on some, and as many elections are marginal, you only have to make small percentages of people vote the way you want to make a difference in a first past the post system. Data manipulation isn’t trying to manipulate your mind, all it wants is your vote, once it’s got that, you can **** off back to your dreary life until next time we need it, prole.

That’s how it works.

FWIW I’ve got relatives in a small town of ~14,000 in Canada, they had an outbreak, and within days had split the population into 3, those that were symptomatic, those that were likely exposed via contact tracing, and the low risk, and tested everyone in that order. So within days you know whether you need to be staying home or not and the town goes back to normal-ish because the outbreak is contained. That’s how it should work.

Sounds good. But how does that scale up to London or Birmingham with millions of people? That’s the challenge.

That’s how it works.

Except, as I mentioned above (and confirmed here) the NHSX app only submits your recent contact data if you are suspected of having C19. So the vast majority of people are unlikely to ever send in details of recent contacts. If you don’t use the app but then test positive the manual contact tracing team will still want to know who you have recently been in close contact with, so will still get a similar data set that will most likely end up in the same place. Will you refuse to work with the manual contact tracing team because of the same privacy concerns?

You all know that it doesn’t actually report on location just on proximity to other users right? The post code is just for a rough idea of where you live.

I think some of you might need to re-read the blog post.

Try again…

Maybe I’m reading it wrong, but I don’t think so.

Sorry – I was reading stuff wrong. It doesn’t work if the iPhone goes to sleep unless another device wakes it up. Situation remains the same. As you were.

You all know that it doesn’t actually report on location just on proximity to other users right? The post code is just for a rough idea of where you live.

…and if you put enough of that information together (along with other bits and pieces) it is likely very possible to get a good idea of your movements without having to directly trace it.

For example people using strava can effectively block where they live from showing up on their feed to prevent bike thefts. The very absence of this data (possibly combined with the data from friends/connections) could be used to determine with a pretty high degree of accuracy where you live.

Even if that were true, are you suggesting this is different for the millions of people who give away this information voluntarily to Google/Facebook/Instagram every minute of every single day as apposed to the Potential of this to do it for a few months to get us out of the shit?

This sort of petty scaremongering gets right on my tits.

It’s not petty scaremongering.

It’s genuine privacy concerns.

Those with privacy concerns don’t give all their stuff to Facebook.  And that’s partly why Facebook put a greater weighting on what other people tell them about you.

It;s the job of the NHS app guys and the Govt. to properly demonstrate and even pass a law if need be to ensure that all data more than 28 days old is deleted, can never be used for anything other than the immediate intended purpose, that there is not a unique ID stored for each installation, etc.

The 2 big phone OS companies even came up with a solution, but that was not chosen because they want more data identifiable to specific installations, and data to be kept forever.

It’s not going on my phone.

It’s not going on my phone.

So back to my earlier question – if you contract C-19 and are called by a manual contact tracing team who will take more or less the same details from you, will you comply?

I’m not trying to defend the app, but it does look like not everyone read the description.

IF it is per the description and that does not change (yes I know) then

it doesn’t track your location

there is nothing held to associate your device to ‘you’ apart from the post code, which you enter yourself

I think that if you had some entry points, ie enough known devices with the associated known ID and where they were over time, and got them to wander about an area for a time, then reported their contacts, you might be able to do something about contact location. BUT you have to be one of a few people inside NHSX to do enable this to happen, and even then, you would only know device contact location, not actual individual device owners, you need to correlate some other data source for that. Periodically changing the IDs like A/G would make this harder still of course.

So its pretty hard to get useful stuff from as it is, Facebook is really much easier, if I wanted to be all big brother I’d look at doing something else.

BUT at the end of the day I’m sure the data does have value for other purposes beyond health. If there was a requirement to delete the data after a period of time people would be a lot happier. Even if only a random snapshot was allowed to be retained for future research purposes.

Interesting that they asked the devs to estimate switching to the A/G API.

‘But we give data to Facebook’ isn’t really a ringing endorsement of a strategy.

Facebook, Google, Amazon. Data behemoths. Unstoppable, they don’t answer to anyone. Now let’s give our Government the same power.

Sounds good. But how does that scale up to London or Birmingham with millions of people? That’s the challenge.

With a decent contact tracing app, relatively simply. You find one person with it, and then test everyone they’ve been in contact with, that find a few more, and so on and so forth. The point is that with a working system you don’t need to lockdown and test the entirety of Birmingham and London, just the ones that have been exposed to it.

That’s why we need an app that works, and by works I include enough people trusting it with their data.