sunnysidedown
caput mortuum
care to elaborate? (if that’s not too condescending)
UK coronavirus tracking app - discussion
- Thread starter editor
- Start date
care to elaborate? (if that’s not too condescending)
mauvais
on reddit or something
The privacy stuff isn't enormously interesting to me, I feel like none of us have enough information to make a judgement either way. I don't distrust the prime of NHSX by default, and I'm sure Apple/Google store approval will mean a careful look at what it does. If you have a smartphone and you're not a weird privacy nerd then you've almost certainly submitted more about yourself already than this app will track. Plus it's a global emergency and it seems like a weird hill to die on. Whatever.
Most of the technically focused commentary (broadly, not just here) has been misguided in some way.
More fundamentally even if it worked I don't see how consumer phone radio contacts (BLE or whatever) are in any way analogous to CV19 transfer. Even if we naively assume that rollout is ubiquitous and Bluetooth works continuously and perfectly records every device it sees in a traceable way, the range on it is too high. Most of the contacts pose no risk. You could get something useful about risk from prolonged contact but you tend to know who you've been in prolonged contact with in a lockdown. There are 'N degrees of Kevin Bacon' edge cases here where it's useful - do you know where your kids have been? - but on the whole it's not valuable information. Then there are all the indirect contacts - someone touched something or you went to a place with air conditioning and you got CV19, but the person you got it from wasn't actually there with you. You might as well do straight location tracking without any of this device visibility shit. Even if you ignore all this and assume it just works somehow, the incubation time is too long that I don't see what you can do with the findings in terms of warning people, other than provide constant warnings to anyone who goes anywhere ever.
IMO this amounts to a tool for 'interesting pattern analysis' rather than having any import to either getting life back to normal or a pragmatic short term medical response.
Most of the technically focused commentary (broadly, not just here) has been misguided in some way.
More fundamentally even if it worked I don't see how consumer phone radio contacts (BLE or whatever) are in any way analogous to CV19 transfer. Even if we naively assume that rollout is ubiquitous and Bluetooth works continuously and perfectly records every device it sees in a traceable way, the range on it is too high. Most of the contacts pose no risk. You could get something useful about risk from prolonged contact but you tend to know who you've been in prolonged contact with in a lockdown. There are 'N degrees of Kevin Bacon' edge cases here where it's useful - do you know where your kids have been? - but on the whole it's not valuable information. Then there are all the indirect contacts - someone touched something or you went to a place with air conditioning and you got CV19, but the person you got it from wasn't actually there with you. You might as well do straight location tracking without any of this device visibility shit. Even if you ignore all this and assume it just works somehow, the incubation time is too long that I don't see what you can do with the findings in terms of warning people, other than provide constant warnings to anyone who goes anywhere ever.
IMO this amounts to a tool for 'interesting pattern analysis' rather than having any import to either getting life back to normal or a pragmatic short term medical response.
souljacker
innit
You were wrong first.
souljacker
innit
The data it logs includes the rssi which should indicate how close you actually got to someone. It also records the length of time that you were close to them. So someone strolling past you on the street wouldn't get the notification if you developed symptoms. In theory.
mauvais
on reddit or something
It is just theory though. Ignoring the many technical and policy barriers to actually doing this, I've done telecoms work in the past (e.g. client network selection policy, cellular vs Wi-Fi) involving signal strength measurement and it is, on a single device basis, garbage data. You could likely crowdsource something better in terms of observations but this use case isn't about crowdsourcing, quite the opposite.
I could be wrong, but I think it's a blunt instrument that inherently lacks the subtlety required to say 'hey you might have been infected by this person you don't actually know'. I don't really know why anyone's taking it very seriously. Maybe as part of very stringent measures it could have some limited utility. That's about it. In your street example, unless everyone takes distancing more seriously and is mandated to wear masks, and cases somehow got to be rarer than NHS PPE, then the person strolling past might be exactly the person who needs a notification.
shifting gears
Well-Known Member
Can some of the more clued-up posters on this thread, who are to varying degrees slating this app/track/trace method, explain why contact tracing has continually been stated as a key factor in the approach of counties that have supposedly managed the virus well, using a similar method?
mauvais
on reddit or something
Possibly because it might work (does it? any actual evidence?) when public cases are extremely rare, the environment is highly controlled and a response is highly organised. We're literally months too late for that.
shifting gears
Well-Known Member
Neatly routes us back to the negligence of those at the wheel, but we knew that already I guess
souljacker
innit
This is Bluetooth though, not cellular or WiFi. I have installed Bluetooth beacons at loads of places and they are pretty accurate. Personally, I think the tech is reasonably sound. It's the app and how it is deployed that concerns me and makes me think it won't be effective.
South Korea seem to have made it a success but they are very advanced technology wise as well as being quite a compliant population. The majority have smart phones and the majority have probably installed the app.
mauvais
on reddit or something
It's the same radio tech. You can't make good decisions from one device's observations of RSSI in a real environment. Again this is to say nothing of the fact that a lot of this information is now prohibited.
Everything I read is along the lines of 'oh it'll probably work' and I find it really odd.
Part-timah
Well-Known Member
You’re assuming this will be effective. NHS IT projects are notorious for being disasters.
neonwilderness
What would Badgers do?
I think they’re relying on lots of people using the app for it to work. But it seems questionable if the app will work in reality as it sound like the app needs to be running in the foreground or near another phone running it in the foreground for the contact tracking to happen.
My understanding of the privacy issue is that they’re using a private key that doesn’t change to record the contacts on a central database. While it’s anonymous (or so they say) it allows them to start building a model of which keys have been in contact with each other and when. If they (or someone else) is then able to associate the private key with an actual person they potentially have a model of people’s interactions which could be a bit Big Brother. Its probably fairly unlikely, but it’s still a concern.
The Apple/Google approach uses a daily unique private key and the contact tracing is done locally on your phone. If someone reports as being infected then their daily keys when they could have been infectious would be added to a central database. Your phone would then periodically check these keys to see if you’ve been in contact with any of them.
So the central database would only have a record of potentially infected keys rather than all interactions.
Plus the app is being developed by a mate of Dominic Cummings, so alarm bells should be ringing anyway.
mauvais
on reddit or something
Stated where? Tested and works in terms of what?
weltweit
Well-Known Member
I think there was a smaller test with military personnel.
mauvais
on reddit or something
I don't think NPfIT was a total failure, but it certainly wasn't a success. In any case there is zero comparison between something of such vast complexity, and a single app, although government's record on software is not exactly great.
mauvais
on reddit or something
Not your problem, but 'it worked' could mean anything. For example an off-market install on a set of Android phones is vastly different to getting something approved by Apple and available in the App Store. Successfully registering contacts in some form is very different to registering useful contacts. Being medically relevant overall is a whole other thing. Even the IOW trial is very, very different to most other places (although it has its own challenges I guess, like how mobile phones don't exist in the perpetual 1950s...)
I guess people really want this to work but I think some serious scepticism is due.
weltweit
Well-Known Member
It seems to me to be a continuation of the mistrust in experts as mentioned by Gove. Who could be more expert in smart-phone apps than Google and Apple, yet no apparently UK NHS/gov knows better, of course they do!! ?
SpookyFrank
A cheap source of teeth for aquarium gravel
In order to work the number of infectious people would need to be below a certain threshold, which is almost certainly a lot lower than where we are now. Contact tracing, where it has been effective, has also been done by actual humans and not ill-conceived apps thrown together on some consultant's lunch break.
DexterTCN
Troy and Abed in the morning
track trace and isolate, I thought that was what was being done in other countries, not just track and trace
mauvais
on reddit or something
This one is complicated. It's not really a question of technical expertise but policy and politics - what Google/Apple will condone on their platforms. Apple have always been restrictive but years ago, at least on Android, anything that was technically possible would have been permissible, but we've since had years of tightening the screws on privacy. Most of the significant updates in new versions of Android are privacy related - things apps can't do any more. I can simultaneously understand why state agencies want better, more centralised and more usable info and why G/A want to maintain that privacy trend.
I guess it comes down to politics and maybe UK gov thinks it can strong arm Google/Apple into allowing whatever they decide to do in the interests of saving lives. It probably is a bit flexible but I wouldn't count on them getting a pass. Remember the FBI went up against Apple and lost.
A contact-tracing app is a great idea in principle, but whilst the present government is in office and there's nothing in place to stop them using data for whatever dodgy purposes they like, I won't be installing anything. I'll consider using it when BoJo, Cummings and the rest of them are in prison. Until then they can fuck right off.
chilango
Hypothetical Wanker
mauvais
on reddit or something
I think that will be a hard sell to be honest. Of all the many ways to blame the public, that's a more difficult one.
I think there are interesting secondary effects that I haven't fully thought about. The idea that you need a bit of modern tech (and likely a specific set of requirements) is an entrenching of an existing digital divide. The idea that government mandated software has a place might set an interesting precedent. The relationship between government and platform providers, as above, will be 'interesting'.
Similar threads
