What I find far more unbelievable is that countries – at least European countries that through the EU have a common legal framework – did not get together and share the bill for the development of a common open source app. Perhaps each country would have to do some minor local customization, but surely the vast majority of the code could be shared?

I frequently find myself pondering this whenever my government announces yet another contract for (the admittedly decent) IT services for interacting with state entities.

In any big organization, the more you try to "share", the more bureaucracy is involved, because you have to get people to agree. You can hardly get agreement on how things should be done in a company, let alone in the EU.

In any big organization, it's better to ask for forgiveness than for permission, meaning that it's better to build it first, prove that it works and then seek to establish it as a standard, than vice versa. And if there are multiple competing proposals, then competition is healthy.

Do all countries have the same commitment to privacy? No. Are all countries willing to work within the limitations imposed by Apple/Google? No. Disagreement is natural and it would be pretty stupid for everyone to stand still until an agreement is reached.

Besides, 20 million euro for a country like Germany is peanuts. All EU countries have spent far more than that during the lockdown.

I think they wouldn't be able to agree wrt even the most basic data privacy questions.

It would actually be quite interesting to compare the different apps and their philosophies (privacy, security, centralization vs. decentralization, open vs. closed source, source code..).

The money isn’t spent on the app; it’s spent on the backend, and there, scaling efficiency gains are limited because of diversity in computer systems.

Also, the app would have to be translated in many more languages. If you don’t want that to increase time to market, it will cost a lot of money.

But in my understanding the backend of the german app is generic, since it only receives random IDs from the mobile apps and assembles a list of them to download again.

How much would you budget for this?

I am actually curious, because on the other hand you would have all the constraints like development in less than 3 months, need for the app to be waterproof in a legal sense (data protection), additional backend development and also development for both iOS and Android.

this is one of those cases where people on here will look through the github repo and say "how could this cost 20 mil?? I could build that..."

maybe they can build a toy version, like everyone else.

The 20 million are also for the backend. I'm not sure it's possible from the public information to determine whether that's a reasonable sum or not.

There’s also a website, there also was a security audit, there’s a hotline for getting codes when the laboratory can’t get you one. There are two separate apps for Android and iOS. Both the website and the Apps are available in German and English with other languages like Turkish following.

All of that costs a lot of money if you want to bring it to this level of polish.

> there’s a hotline for getting codes when the laboratory can’t get you one

This cannot be overstated and is generally underestimated by developers. Setting up a hotline infrastructure (actually it's two hotlines, one which is more like a helpdesk for customers of the app and one for the lab stuff) and operating it produces significant initial and ongoing cost.

Also they will still have additional ongoing cost for infrastructure (server operation, traffic) and cost for support regarding the still-to-be-performed process of interfacing the lab information systems of the 140-or-so labs in Germany with the Corona App Backend infrastructure (which is necessary to allow for the lab hotline and TAN stuff to become replaced by direct transmission of test results to the user via the app and its backend).

If those operating costs even just for the near/foreseeable future are included in the 20M€, I consider that a fair sum.

Up here in DK, 20 million EUR is approximately the price of 200 manyears of labor, with taxes. Assuming everyone is as highly paid as a senior developer/tech lead/managerial position.

The efforts are definitely not free, but the price is entirely disconnected from the service provided. It is just the result of the usual enterprise companies that handle government contracts.

Such grotesque waste of tax-payer money should be made illegal, but the fact that these companies keep getting contracts is unfortunately reinforced by absurd government processes.

how much do you think an end-to-end product/service like this should cost

To come up with a clean-slate estimate would require quite a bit of work. Instead, we can walk backwards from the data we have to price the actual work. Anything below is just an out-of-my-ass guess based on experience on certain projects, experience with security validation, and what information is easily available.

If we start just with development, we see that the the Github organization has 30 members. Unfortunately, the companies had absolutely horrendous source control and open source workflow practices (e.g. the android app's first commit is 35k lines), so we don't have a good history of all contributors, so for good measure let's double that to get an absurdly large team of 60 developers.

Considering the pandemic, let's say they all started working on the project in January. That means that the cost of the software developers themselves is approx 30 manyears, or 3 million EUR.

Translated the limited content and creating the webpages are not very expensive, but let's allocate a whopping ~100k EUR for that.

We need negotiation, design and security validation as well. Let's give that 5 people, 2 months at 4 times the already expensive senior developer salary. Expensive as hell, for no good reason, apart from to cover absurdities. ~200k EUR.

All these numbers are absurdly high (i.e. room for large errors), and we're "only" at 3.3 million EUR in cost. Need to add some profit margin for the company, but with more data (e.g. actual developers it took, how long they worked) the estimate could probably be pushed much lower.

They also load tested it with “two digit millions of users” (https://www.golem.de/news/bundesregierung-entwicklung-von-co...)

Depending on how realistically they did that, that can be costly.

If this system stays up on its first day of use, 20 million might be worth it.

The hotline will cost additional 2-3 million each month!

I‘ve seen larger amounts spent on startups with no significant userbase nor exceptionally well made products.

Unbelievable for regular software maybe. Quite believable for enterprise contracts, downright cheap for healthcare stuff.

Jens Spahn (M-of-Health) said yesterday in Tagesschau that the app was within budget and delivered on time, which he admitted himself is rare!

Actually, it's easy to deliver something within an overblown budget.

But also rare ;)

Additionally, there are expected operating costs of 2.5-3.5M€ per month. Most of that will go to the telephone hotlines.

Really? Maybe I'm just too jaded but I feel like 20 million is not a particularly crazy amount for a government project, especially one that is this important and had to be developed under this much time pressure.

Compared to the billions to potentially trillions of euros Germany will be paying for the lockdown and the other EU states this is basically nothing I think.

Provided the app will solve all those problems single-handedly. There are voices arguing the impact will be about 0.

Are people really bickering about the price of this? Please tell me you're just German developers angry you didn't get this contract.

Because the alternative is that you're terrible at math.

According to[0], the costs of the pandemic for the German economy are from 255 to 729 billion euros. Taking the middle of 500 billion as a best guess, and a quick division will tell us that this app will be worth it if it stops one in every 25,000 infections.

Or, alternatively: this app costs about 0.25€ for every citizen, or about a quarter of one mask. That does not strike me as a terrible deal.

The point here isn't that this is a model for how all government procurement should be run. It is not.

It's that under the specific circumstances, overpaying is completely acceptable if it meant, for example, getting a contract signed a day earlier. Note that situation was obviously far worse when this project started than it is now.

[0]: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&c...

You're right of course. My point was rather that you cannot simply say it's worth a single effort because the total net benefit of all measures is positive. Thanks for doing the math. Quite interesting result.

The worst case is that this app has negligible impact and even if it did, trying it as a strategy that helps fight Corona is still worth €20m.

Paying students €500 a month as emergency help will cost hundreds of millions and this is a single miniscule effort in a package of several hundred BILLION Euros.

Discussing the cost of this app with the given circumstances is a complete waste of time. It's a literal rounding error in the overall Corona bill for Germany alone.

Ah, then I may have misunderstood the sentiment. In that case, please excuse the (passive-?)aggressive tone of my post.

And yes, the link is just the result of some hyperactive copy&paste.

I'll now try to enjoy the thought that the Germans are bickering about the price of this on HN. It's sort of a return to normalcy: the luxury of caring about pointless details precisely because the government isn't quite as incompetent as always assumed, at least compared to some others.

If you assume a linear relationship between economic damage and number of infections.

This! The economic damage is largely caused by our response to the infection risk, rather than by the infections themselves. There are numerous countries with enormous economic damage and very few infections, and others with large numbers of infections and relatively less economic damage. They don't seem to be particularly correlated.

Completely OT and just out of curiosity: Is there a point to posting the source as a redirect via Google? Or is it simply the link copied from the search results page?

Germany has done better than most countries, and we need to throw everything at the problem including the kitchen sink. This is noy time for meanongless bickering, its the time to enable folks with a plan and then access them on their merits.

The well of criticism is endless. There are voices arguing that masks are useless, that 'Churchill wouldn't wear one' and that 'social distancing is communism'. Dealing with all of them will result in brain damage.

It will absolutely be 0, I didn't mean to imply otherwise.

20 million is still 20 million

For a system which interacts with Germany's divers structure of health authorities. Last time I counted there were 374 municipal health authorities, all of them have to be integrated. Also the laboratories and doctors. In addition they set up two calling centers (a technical hotline and a health hotline) and did multiple audits for security and for epidimological sense. And then all of it, frontend and backend, is open source.

For a government project that is cheap! (25cent per citizen)

Edit: source for the 374 is the table on https://de.m.wikipedia.org/wiki/Gesundheitsamt

The Federal Government also sourced PPE for around 900 M € without really distrbuting it. Also, without paying suppliers on time, or at all. So I guess the app is kind of cheap?

Well, if they didn't pay the suppliers, they didn't really spend that €900m, did they?

They sold at fixed prices without limiting quantity or specifying requirements. They are already getting sued over it.

Edit: 600 million, source: https://www.spiegel.de/politik/deutschland/corona-krise-jens...

For comparison, the Italian version was donated to the government by the developer, Bending Spoon. They are not small, but I really don't think they would have been able to donate 20M€ of effort.

To be fair the backend, while developed by Bending Spoon, is operated by the government, while the 20M€ from the German government might also include operation? But still

According to the git log the overall development time was round about 6 weeks. Including some holidays in between, we're looking roughly at 1 Million Euro per development day!

I suspect that you don't have the slightest clue about the effort that has gone into it. This morning, Timotheus Höttges talked a little about that. [1]

Sooner or later there will probably come auto-translated captions.

[1] https://www.youtube.com/watch?v=GDnaNMGbXnw&t=1667

20m€ are peanuts for this. These government contracts normally takes ages, explode in cost and never really work in the end. I'm glad to see that we have experts in germany that can in just 6 weeks develop such an app for just 20m€. I don't get why people think it's too much. It sounds large but we're not talking about your everyday startup here, that is half-assing their product for 2 years and have tons of bugs in it. This app seems solid and usable for 80m people