Ireland donates contact tracing app to Linux Foundation • Hacker News Board

Re: Ireland donates contact tracing app to Linux Foundation

#111

Post by ericlewis » Mon, Jul 20, 2020, 10:48 PM UTC

Someone wrote: post #10

Sephiroth87 wrote: post #4

Earlier quoted context omitted.

I wouldn't put the focus on it being ONLY 850k, that seems a competent price, I would really like to know how in the world they spent 11M on the uk one...

I don’t know specifics about the UK version, but most of those costs likely weren’t on the app proper. They did run a field test, for example. For comparison, the German version cost millions, too. That included multi-language support, setting up help lines, security evaluation, load testing millions of concurrent users and, IIRC, connecting to quite a few of different hospital computer systems. Add in time pressure,…

We ran a field test too. But again; I probably can’t talk about that much. Sorry. If you have more in depth technical questions I might be able to answer them.

Re: Ireland donates contact tracing app to Linux Foundation

#112

Post by ericlewis » Mon, Jul 20, 2020, 10:49 PM UTC

donaltroddyn wrote: post #19

Someone wrote: post #10

Earlier quoted context omitted.

I don’t know specifics about the UK version, but most of those costs likely weren’t on the app proper. They did run a field test, for example. For comparison, the German version cost millions, too. That included multi-language support, setting up help lines, security evaluation, load testing millions of concurrent users and, IIRC, connecting to quite a few of different hospital computer systems. Add in time pressure,…

It also includes the DPIA, which in itself can be an expensive process. There was a lot of concern/press in Ireland pre-release about the privacy implications of the app. It's hot on the heels of a "mandatory but not compulsory" national ID card fiasco that was determined to be unlawful by the privacy regulator last year: https://www.thejournal.ie/psc-no-legal-basis-4766822-Aug2019...

You’re not wrong.

Re: Ireland donates contact tracing app to Linux Foundation

#113

Post by ericlewis » Mon, Jul 20, 2020, 10:51 PM UTC

addicted wrote: post #9

Sephiroth87 wrote: post #4

Earlier quoted context omitted.

I wouldn't put the focus on it being ONLY 850k, that seems a competent price, I would really like to know how in the world they spent 11M on the uk one...

That’s like 4 SV devs working a year. And just in salaries, not including payroll taxes, office space, admins, QAs, etc. 850k sounds eminently reasonable for such an undertaking in such a short period of time.

We had a team of great devs and designers who worked on this. What most would consider “par for the course” in relation to an MVP startup for instance.

Re: Ireland donates contact tracing app to Linux Foundation

#114

Post by quonn » Mon, Jul 20, 2020, 10:52 PM UTC

johannes1234321 wrote: post #83

rkachowski wrote: post #76

Earlier quoted context omitted.

In what sense are most of the bugs from the "system framework"?

There is this contact tracing framework in iOS and Google Play Services and that causes a few errors. Some users get a message like "not possible in your region" and some get some random error codes about rate limits and such errors, which bubble up from those frameworks and can only be fixed there.

Yeah. It‘s the framework which does pretty much all of the interesting technical work, if I may say so. It does the communication with other phones and determines if there was a contact and generates, stores and maintains the tokens and given the server data figures out matches and so on. It has a tiny, simple to use API on top.

It was jointly developed by Apple and Google. For free.

Re: Ireland donates contact tracing app to Linux Foundation

#115

Post by dwheeler » Mon, Jul 20, 2020, 11:10 PM UTC

tsbinz wrote: post #33

cj wrote: post #22

Where has a contact tracing app been successfully deployed at a large scale? Have those places seen any positive effect? Apple/Google's APIs were widely discussed a couple months ago, but (at least in the US) it seems to have fallen off everyone's radar.

The problem with contact tracing apps is that their effectiveness has a quadratic relation with the adoption rate. They would be mostly useful to trace "random" contacts (since you tend to know which of your friends and family you saw within the last 2 weeks), but to be able to trace a possible infection both the infected and the "recipient" have to have the app and active. In Germany they said their 12M installs is…

I completely agree with you that this cannot possibly replace other measures such as masks, hygiene, and avoiding large indoor gatherings. Nor should it.

I think there is an additional confusion though. This is really an exposure notification app. If you used only this, and not traditional contact tracing, you would have to have a lot of adoption. But I think the general recommendation is to use both (1) these exposure notification applications and (2) traditional contact tracing by people trained to do it. Then you don't need as large an adoption to get utility. Traditional contact tracing can be effective, especially for people you know you contacted. But it is slow, and does not handle well the people you don't know. By combining them, there is a greater odd of notifying those who may be potentially infected. As long as you're also doing traditional contact tracing and other measures, a lower adoption rate for these applications can still be valuable.

Full disclosure, I work for the Linux Foundation. But I still believe this anyway :-).

Re: Ireland donates contact tracing app to Linux Foundation

#116

Post by Polylactic_acid » Mon, Jul 20, 2020, 11:28 PM UTC

edaemon wrote: post #35

cj wrote: post #22

Where has a contact tracing app been successfully deployed at a large scale? Have those places seen any positive effect? Apple/Google's APIs were widely discussed a couple months ago, but (at least in the US) it seems to have fallen off everyone's radar.

South Korea has nationwide contact tracing that's been quite successful. Here's a study with some data on it: https://wwwnc.cdc.gov/eid/article/26/10/20-1315_article New Yorker article that's older but adds some color: https://www.newyorker.com/news/news-desk/seouls-radical-expe...

Australia has had an app for a few months now that has not provided any useful information since its creation.

Re: Ireland donates contact tracing app to Linux Foundation

#117

Post by quonn » Mon, Jul 20, 2020, 11:32 PM UTC

oezi wrote: post #110

markdown wrote: post #92

Earlier quoted context omitted.

$10M for two mobile apps, some server side code, and docs? And you're defending this as if it's reasonable? One could build 100 apps with server code and docs with that money, and still have a few million left over for hookers and blow. Stop defending corruption.

I honestly think this will be a net loss financially for SAP and Telekom. To get it done so quickly in parallel to the contact tracing API being just developed will have required a team of their best developers. The lost opportunity cost for this alone will be huge. Then it is not just development, but also QA, community management (git hub issues are in the hundreds), legal, project management, etc. I think the prim…

I do work in a corporate setting and have done so, as a consultant, for years. I know the numbers very well, both from what I and others charge and what large companies actually budget.

Re: Ireland donates contact tracing app to Linux Foundation

#118

Post by stirlo » Tue, Jul 21, 2020, 12:27 AM UTC

Polylactic_acid wrote: post #116

edaemon wrote: post #35

Earlier quoted context omitted.

South Korea has nationwide contact tracing that's been quite successful. Here's a study with some data on it: https://wwwnc.cdc.gov/eid/article/26/10/20-1315_article New Yorker article that's older but adds some color: https://www.newyorker.com/news/news-desk/seouls-radical-expe...

Australia has had an app for a few months now that has not provided any useful information since its creation.

The Australian app (COVIDsafe) does not use the Apple/Google API becuase the Goverment wanted location tracking which is banned under their terms.

This means the COVIDsafe app chews up battery life and is vulnerable to power saving features which shut down its communication after a short amount of time on iOS in particular.

It would have been far better to use the API and obtain the location data from elsewhere (Google, Ad networks, Mobile Carriers) if really needed. If this were a less free country I'd suspect the location tracking was part of a government spying program but instead I just chalk it up to incompetence.

Re: Ireland donates contact tracing app to Linux Foundation

#119

Post by markdown » Tue, Jul 21, 2020, 2:13 AM UTC

tester756 wrote: post #99

markdown wrote: post #92

Earlier quoted context omitted.

$10M for two mobile apps, some server side code, and docs? And you're defending this as if it's reasonable? One could build 100 apps with server code and docs with that money, and still have a few million left over for hookers and blow. Stop defending corruption.

>One could build 100 apps with server code and docs with that money, and still have a few million left over for hookers and blow. 100 random apps or app that have to handle pandemic in country? Also what about support and infrastructure to handle those milions of concurrent users

> 100 random apps or app that have to handle pandemic in country?

The contract tracing apps are really basic... a step above torch apps.

Re: Ireland donates contact tracing app to Linux Foundation

#120

Post by patentatt » Tue, Jul 21, 2020, 2:22 AM UTC

tlarkworthy wrote: post #40

quonn wrote: post #34

Earlier quoted context omitted.

I said a few 1000 and I actually browsed the code on Github when it came out. There were just six weeks to develop it. And the concepts where developed before by public research institutes for a tiny fraction of that price (not included). (If you insist, here is a small calculation. Let‘s say we charge a whopping 500 Euros per hour per engineer. In other words an engineer that costs 1 million per year. And we have 6…

20M for the boost of Germany's worldwide status is a drop in the ocean. The money was well spent, Germany did not embarrass themselves by penny pinching. It needed to work, and it appears to. Compare this to the UK who is still flapping around.

Yeah, I don’t get all these people criticizing the cost. Also, when a government is dealing with a flagship corporation of their country, cost is a secondary or tertiary concern anyways. It’s a subsidy in part, and it’s also just that $10m for a nation state government to get it right the first time is just so so so inexpensive. I think SAP let them off easy this time, if anything.