 Yeah, and this ties in nicely with what Thomas just presented. Also, another means of testing our software on users, just without the annoying talk of the part, right? So this is essentially a follow up of what I presented last year. The user feedback framework that allows you to gather some of the trick data in your application. There was some interest in this last year, but there was broad agreement that we would need some kind of rules to govern how we would use this. In particular, quite simple. Since this is one of our unique selling points, we plan on allowing that to be discussed when we're importing anything like this do. So we had an extensive discussion and I've been doing it for about a month. And the result is the KDE-10 entry policy that's in the community. When I worked the abstract, I talked the discussion was the wrong one, so that's why I said it's about the open questions that we are in anymore. So I just show you the key aspects that are in that policy. One is transparency. So we show, or if you use the termetry, you might be required to start to show very explicitly in detail what data is submitted related to the user's understanding and it's necessary to make up what data is sent and so on and so forth. So that the user can make an informed decision or they're against this or not that's going to come up. The second thing is about control. That advice is pretty obvious. We never said anything without the new lens. Specifically, you have the option to opt out again at any point in time and it's not tied through any certain features or sending an email to the individual and trying to do a telemetry thing. That's all right. And then the next part is about the identification. I think that's going to add some of the most intense discussion because the user system in use by testing will optionally use unique identifiers. And the policy that's coming out explicitly prevents the use of any kind of unique identifiers. And that has the convenience side effect that we have a lot of talking about personal data because it's not tied to a person. Like somebody is using Plasma 512 in a French local and the system is relatively often like that's not personal data so we're not far off or we're not affected by all the personal data regulations which makes a whole lot of information inside because it works easier. This has to be a rotating infrastructure that is also part of the policy and another somewhat more emotional side is that we want to have people who don't have quite access to this and basically anyone in the community should have access to the data and have access to this and going away. That goes together with the unique identification part of the outbreak. In its own personal data we know we restrict access control in the data but ideally if the data really is completely anonymous and also no correlation actually between people it could even publish it. So the idea is to make the data really just data and of course the data so in an extreme case we don't really need any access control. I'm only at 5 minutes in total so that's surprising. So the user feedback also got changed based on that discussion. Some of the data was integrated there but it's important to remember that the user feedback, the framework and the digital data policy are more or less integrated. The policy doesn't act as the use of a specific framework nor does use of the framework or has to be complied with the policy. The framework is very flexible in view of the data source system by the policy and of course there's nothing wrong with it that provides with the policy. Nevertheless we added a few more transparency in the digital feedback not just showing you the data but the entire data that was ever submitted in the past to give you a more precise view on what this means. We added a lot of kill switch so just to give you that that's why essentially that's important if you do study a more important application and the effect of the operation that it will serve as a sub-example actually deploying this. So yeah, yeah I think now we're finally in the position to actually deploy this on a digital structure every time the application just took one extra year but unless somebody finds more issues with this I think we actually start.