I completely agree - I came into it against the centralised model on principle but after reading their whitepaper I actually think the benefits may slightly outweigh the negatives.

The app has to actually work for a high enough proportion fo cases, though.

Centralized models would be more palatable if the keys to the kingdom were stored with an NGO.

There's too much temptation for governments with poor privacy track records (and that's where they're even trying) to stick their hands in the cookie jar for other purposes.

If the government wants the data, an NGO offers almost no protection. The only safe choice is not to collect the data in the first place.

An NGO offers substantially more protection than a direct government store.

Whether that is enough protection is a fair debate.

And for it to work people need to trust it.

True, and I would not trust it if I had a phone that could run it. It's not trust in the app but in the government.

I recall some dickhead politician being asked about security of a centralised app, his response was "GDPR would take care of that". Because hackers care about that kind of thing.

Honestly this uk government is so bloody inept. When you want theresa may back, that tells you something.

The explanation of the trade offs by Ian Levy (Technical Director of the UK National Cyber Security Centre, not the UK MP with the same name) was actually very well put, but few newspapers over here in the UK even seem to have read it or deliberately ignored it.

https://www.ncsc.gov.uk/blog-post/security-behind-nhs-contac...

I think part of the problem with it is that the beginning is a bit of a long winded intro, but the technical details later are really good.

Apparently the particular decentralized API exposed by Google and Apple might have some limitations in terms of being able to accurately estimate things like the distance between people compared to using Bluetooth directly too, which is a bit of an issue: https://www.bbc.co.uk/news/technology-53095336

That was the excuse given by Matt Hancock but it's almost certainly bollocks.

Google has 7,000 people working on maps, Apple has 4,000 people working on maps, is it really possible that a few app developers assembled by the tories (who have a track record of failure and lies) have stumbled upon a more accurate way to calculate the distance between devices than the 11,000 people that work on these problems full-time have ever managed?

From the article: > The centralized approach would allow much more data analysis

If this was really a benefit, I'd expect the journalist would have been briefed on the beneficial outcome of the data analysis, rather than just "more data analysis".

There have been quite a few articles on the benefits. For example the central model would allow the NHS to quickly spot a spike in cases in Town X - a pretty important benefit.

The independent Covid Symptom Study app from Kings College, already on 3 million UK phones since March, took three days to write and largely would do that if it was more publicised by the government.

The NHSX app tried to do too much — trying to collect centralised data when one key function didn't need to be centralised was always going to be an issue when trying to get people to install it in large numbers.

Smaller samples are enough for population surveys, while high percentages are needed for the track and trace function. Mixing up the two into one app has doomed this project.

> a centralized model based on confirmed cases rather than suspected ones was more in line with capacity [of testing and contact tracers]

Seems a pretty clearly spelt out to me.

I'm... not sure what that means?

Doesn't it mean: there just aren't enough tests for this to be worthwhile?

[Edited to add: I re-read the article and - yes, it means there is not enough testing capacity. Limiting case detection to the number of tests you have available is absolutely not a benefit, unless you are a government trying to obscure the truth, perhaps]

No, you have still misunderstood. The NHS model collects more data than the google/apple one. The idea was that that additional data would allow limited testing capacity to be used more efficiently. For example, false positives for outbreaks would easier to detect, as well as wider trends.

Also, it's not just tests that are short, but contact tracing resources. Basically, when you shut down a business or location due to an outbreak, you want to be as precise and accurate as possible - to do otherwise will cost the affected too much, and they'll start to ignore the contact tracing system entirely.

We'll probably never know now if there was an actual benefit to the NHS's proposal, but it seemed at least plausible.