Ireland donates contact tracing app to Linux Foundation
nearform.com
nearform.com
So the total cost will be 50-100 millions. Despite the fact that most technical work was already done by Apple and Google for free.
I‘m not even talking about the fact that my dad got a new phone to run it - sold by T-Mobile (Telekom) of course.
And hardly anyone complained.
Plus, I just don't understand all the NIH. Why isn't this kind of stuff being developed in one place and then forked by the respective countries instead of everyone creating their own tiny island?
That said, I don't have any major gripes with how the code turned out, just with the cost.
I have followed this project closely. Checked the protocols. The APIs. And what‘s built on top.
The only complex thing I can think of is the way cases are confirmed. That‘s a bit involved. But it was worked out in advance.
When I checked the actual Swift code was just a bit less than 10k.
The backend apps are fairly small Spring boot apps.
There was a messy attempt before (not finished in time at all) and they fixed and finished that attempt.
It boggles my mind that I have to remind many app publishers that we live in a globalized world and that it’s a natter of a few minutes work to just select the countries to make the app available in. More so for free apps where the publisher doesn’t have to worry about taxes and payments.
Even for apps that are not localized to English or other languages, the publishers should prefer to get them out to as many people as possible in the whole wide world who may be visiting or on a long stay in a specific country.
In this case, the app developer can surely include help text that explains where the app is meant to be used or would be useful (it’s also not like people with a German App Store account living in some other country would be able to use this app effectively).
This boggles my mind. Sure, it will take a reasonable effort to get widespread use of these apps. But the impact would be huge. Even low rates of app use have a real chance of lowering the reproduction rate significantly, and reasonably broad usage bring it less than 1.0.
The technology is designed in the right way, where anonymous data is being collected with no app required. People can download the app right after they get a positive test and report themselves positive retroactively. Same on the other side, you can download the app and know how much contact you've had with positive cases retroactively.
I'm no longer surprised the US federal government is doing nothing to push this. But I expected more out of the states.
[0] https://9to5mac.com/2020/07/13/covid-19-exposure-notificatio...
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.
The government was then talking it up with "If enough people download the app, we'll let you go to football games again!!!"
Current media reports say none of our local health departments have found _any_ contacts using it that they hadn't already found via manual contact tracing.
https://www.abc.net.au/news/2020-07-15/coronavirus-app-limit...
(I think 100 apps+backends for $10mil is a bit optimistic, if those backends are required to be HA and support a significant percentage of a countries population. As a single datapoint, where I work has many iOS+Android+backend app projects for gov departments that serve tens to hundreds of thousands of users that cost them high 5 or low 6 figures plus 4 figures a month support/hosting, but if they wanted me to support the 60% of the population (~15mil users) that our CovidSafe app was targeting it'd most likely come to more like mid to high 6 figures, and certainly 5 figures a month to host/support it. So I'd have said maybe 20 apps realistically for that price. Still just as fraudulent a rorting of taxpayer money though...)
The loss in public reputation if they had failed was not worth even participating.
Also, having worked in a corporate setting would have told you that you can't launch anything for less than 1m EUR/USD, can't launch two things for less than 5m and 3 for less than 20m.
20m really is a laughable sum.
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
Same for what you call concurrent users. Those users have no identity at all and essentially need to anonymously download a bunch of static files which are the same for everyone from a bucket periodically. Put a CDN in front. Done.
The only backend touches that apps do is a few login/register APIs when you first launch it, and if you ever test positive and agree to send your data it sends a big json blob to an AWS Lambda nodejs function that pretty much just uploads the json to S3.
I don't recall how complex the login/register stuff is, but I don't remember thinking "That's gonna need heroics to register 10 or 20 million users quickly" when reviewing the code.
(If you wanna go look yourself, try here: https://github.com/opentrace-community )
The contract tracing apps are really basic... a step above torch apps.
(And the MSS/GRU/NSA/Mossad will all insert their code into your app bundles without you needing to do anything...)
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 weeks time which was all they had. And let’s take 20 people for that. And let‘s add 500.000 for some overhead. Then that‘s still just about 3 million. And I‘ve used insane numbers for a product that trivial.)
Secondly you are only taking in "development work", but such a contract entails other things - certainly there is a penalty if they deliver too late - there is legal risk to cover, if they deliver crap in some legal sense they are liable - probably SAP had other contracts which they had to postpone - count in all the lawyers writing the contract :-D
And yes, they certainly made a profit out of this and yes a proper biding procedure might be better, but given the timelines and the transparency I think it's ok.
Let‘s assume another million for lawyers „writing the contract“. It‘s still far off.
Additionally let‘s remember the 20 million is just for the development. It does not include running the thing. That‘s what the other 50+ million are for.
I don‘t agree. I run a business myself. I charge plenty. I know how much DAX size companies charge as well. The numbers don‘t add up, and this is especially true for something that could have been done pro bono by SAP in the current situation where pretty much everyone tries to do whatever they can.
- 9.5M development (SAP)
- 7.8M setup/operations (Telekom)
- 2M Support(SAP)
+ Tax
So, in the books, development is 10M, not 20M
(german source, un-paywalled: https://outline.com/UYqb5b)
... celebratory cocaine for the entire sales team, new yacht bonuses for the entire executive team, another few mil into the CTO's golden handshake fund, and a brown paper bag full of unmarked notes for the gov procurement guy. Poor bastards probably barely broke even on the project - certainly not enough to pay out any bonuses to the devs or project managers...
Additionally, for that price you could have easily have 3 independent teams develop it and take the best one and still be cheaper and better.
I hope someone will challenge this in court. I don‘t think they can get away without a call for bids which should have been done (across Europe, as the law requires).
It was jointly developed by Apple and Google. For free.
There's some kind of cognitive dissonance around your idea that these companies deserve so much compensation for designing a solution that relies so heavily (entirely) on these features, and yet be free from responsibility for serious bugs that arise from this decision.
And frankly nobody thinks "wow, Germany has a working COVID app" since almost everyone else has it too. Even Italy, amongst a ton of screw ups, has managed to produce a perfectly fine open source contact tracing app.
US, UK, Brazil, South Africa, Mexico, Chile, Belgium, Sweden are significantly affected countries without one.
I don't expect Germany to get a wow, only abscense of negative press.
Small exchange from the British Parliament: https://youtu.be/atAy8NGOoiw
An illegal subsidy. Public contracts require a bidding process, usually across Europe.
It‘s unfair that smaller companies don‘t get a chance.
I reject the idea that Germany should prop up its biggest companies, excluding what really is the backbone of the German economy (which works very differently from the US one) and also excluding our partners across all of Europe.
I could do that in a weekend! http://danluu.com/sounds-easy/
Not someone taking over a fairly small project, finishing it in 6 weeks and then charging 20 millions for the job.
Obviously, it would have been better if the German app has been cheaper. But rush jobs are expensive, and the cost involved is pretty minimal relative to the cost of managing the pandemic writ large. If it's even slightly effective, it was worth it.
And if the both had failed their reputation would have suffered a lot. Well it's not the best as it is, so well.
They have a top team and delivered on time. Thats rare in software development.
And 20 million are nothing compared to the tax break which costs about 3 trillion.
For the government it's money well spend in two german flagship companies.
My guess most of it was insurance money in case something went wrong, so these companies would have taken the fall.
That‘s like the cost of 1-2 weeks of testing.
But it‘s silly to argue that just because something necessary costs a lot we can disregard all other costs.
For 50 million you can build a school.
Sure, Germany can afford it. But the cost is ridiculously high.
From everything I've read, this kind of app 1) is only really useful if enough people actively use it, and 2) have serious issues with false positives because of inherent positional inaccuracy (or inability to be accurate without sucking battery).
In regards to 1, in my country at least, unless using the app would be mandatory, I know many more people who will refuse to use it or be too lazy to use it. And I'm pretty sure hell will freeze over before the government can make usage (and owning a smartphone) mandatory.
These apps would, granted, be far less useful in places with major ongoing outbreaks and inadequate testing infrastructure; lockdowns should be used to bring numbers down to the level where this sort of thing is workable.
(1) is probably a greater problem. Ireland has about 30% uptake, which probably isn't enough.
Relative to other COVID-related economic damage this is peanuts. It's acceptable to throw a few tens of millions at a wall and hope that something sticks.
> The German government says its app cost 20 million euros ($22.7 million) to develop and will require 2.5 million to 3.5 million euros per month to operate. It’s available in German and English, with Turkish and other languages to follow.
Legally I believe the government is required to answer that, even though it already seems they're trying to weasel themselves out of it.
Why was it canceled?
Research institutes are typically funded through contract grants, where you as a principle investigator are supposed to be using the money for a research project. Obviously there are exceptions, but if you want to make a large name for yourself in academia, doing maintenance projects isn't the way to get yourself promoted; all the flashy rewards are in publishing novel cutting edge ideas in high ranking journal.
I am not an expert on how things are in Germany, but if they're anything similar to the US experience, I'm not surprised at all that they would fail due to the misaligned incentives.
The company is part of the development. Afaik many apps are open source. "Donating it" means offloading the support and profit from Linux Foundations status.
Nearform is involved in integrating the backend into the countries health system... so they offload the app and just do the work that gets paid well. ?
The app has been open sourced by the Irish Government, not Nearform who were the developers. The copyright is owned by the HSE (the Irish public health authority).
The app being open source does benefit Nearform, though, as they can sell their services customising the app for other governments.
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.
There has also been successful deployments in Taiwan and Singapore, but AIUI they don't use the Apple/Google API and are far more intrusive.
[1]: https://de.statista.com/statistik/daten/studie/1125951/umfra...
How many?
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...
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.
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 :-).
I released a blog post (10min read) this past weekend breaking down digital contact tracing in the U.S. I also discussed why Google and Apple's solution isn't sufficient.
https://tolusnotes.com/state-of-digital-contact-tracing-in-t...
https://developer.apple.com/contact/request/download/Exposur... gives you that reason:
“In order to use the Exposure Notification APIs, You must be a government entity, such as a government health services organization, or a developer who has been endorsed and approved by a government entity to develop an Application on behalf of a government for COVID-19 response efforts. Entitlement Profile(s) are limited to one (1) Application per country unless the country has a regional approach, or as otherwise agreed by Apple.”
Of course an app is not a panacea but as developers we can't create vaccines or provide medical assistance. Its not perfect, as it depends on getting enough people to use it. No mean feat when there are people who don't even believe Covid exists or think it is only dangerous to older people and don't want to take any precautions like wear masks or socially distance. Even then, there are lots of technical challenges trying to use a technology like Bluetooth for a purpose for which it was never intended.
There will be a vaccine at some point (plenty of anti-vaccine people too, which might be a problem for the future) and there will be better treatments in the medium term but right now, speed is important and if it helps States to save time and get a contact tracing app quickly then it has to be a good thing. That way we are all safer.
Even despite having milions downloaded copies, then it wasn't "broadcasting" a lot of keys
Can someone say something about this? how valid this on?
In contrast, our nearest neighbour scrapped their app last month after spending over £11MM (~€12MM) and are now rebuilding on the Apple/Google API: https://www.bbc.com/news/technology-53095336
The code is already up on GitHub: https://github.com/HSEIreland
edit: I guess what can be said was it was a highly efficient project, and there weren't so many fingers in the development pot so to say.
Edit: the github history here is what probably happened after they cleaned it up and open sourced it. Which isn’t uncommon I think.
> I am not sure how much I can talk about that
it genuinely seemed like it might be a way to list the problems without saying that there were problems.
I'm sorry that it seemed like I'm attacking the GP. I did not intend to, but I also knowingly ignored this interpretation, because wording is hard
As for problems, not really any problems. It was quite straight forward. I’m just not sure my company would appreciate me speaking on their behalf about details.
Edit: but I can answer like, general technical details and things like that.
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, and costs go up. Germany did get a quality system, judging by what the spokesman for the Chaos Computer Club said (paraphrasing: ‘we have the unfamiliar problem that there is so little to complain about’. They never endorse stuff, so that’s about as endorsing as their statements can get)
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.
EDIT: added the 'with compatabile devices' disclaimer
Meanwhile, Play Store says India's Covid tracing app was downloaded by more than 100 million+ people.
https://play.google.com/store/apps/details?id=nic.goi.aarogy...
I would say the majority of people in Ireland know the various social network apps are a bajillion times more intrusive than the covid checker. Im fairly sure most people "trust" the government not to be "spying" on us
That sounds strange. Explain more?
So, the citizens' choice isn't between one app and another app. It is only whether or not to use the app that has been chosen for them.
Therefore, their choice doesn't tell you much about the quality of the app. The primary driver of the decision is whether or not they're willing to take steps to fight the disease.
It's similar to being served dinner on an airplane with only one meal option. They bring you the food, and you either eat it or you don't. The choice doesn't tell you much about whether the food was good. It's more of a reflection of whether you were hungry. The only thing it tells you about food quality is that it (if you eat it) it was somewhere between barely acceptable and excellent.
Here in Australia, our gov is extremely incompetent at delivering successful IT projects. They also have a terrible track record with securing data, scaling services (eg national-level IT outages), don't admit to mistakes, etc. :(
> It's more of a reflection of whether you were hungry.
No. To go with your example, if you know the people preparing the meal are incompetent, don't follow anything like reasonable hygiene standards, have previously caused significant outbreaks of disease themselves, and each time they're caught they promise to do better (but actually don't do so)... would you eat something they've prepared for you? Especially prepared at short notice where corner-cutting is expected? ;)
Trying to say that only idiots (etc) wouldn't install such an application is seriously wrong. ;)