 Efallai, rydw i'n gweithio. Rydw i'n gweithio, Chris Brind, a rydw i'n gweithio ar y ddechrau ddechrau'r Lapps yng Nghymru. A rydw i'n gweithio i chi'n gweithio ar y ddechrau ddechrau ac mae'r fawr oedd yn gwneud y dyma. Rydw i'n gweithio y ddechrau o'r ddechrau online. Dychrau ddechrau, rydw i'n gweithio'r ddechrau ddechrau'r ddechrau a bryd y'r gweithio'n rydyn ni yn gweithio'r ff vinolau wasanau ac yn gweithio'r ddechrau'r ddechrau. Mae'r ddechrau, wrth gwrs, nid o'n ddench chi'n fawr o ffinsu ar ddechrau eflwyno'n gweithio. Rydw i'n gyffwilio'n ddechrau i'n gweithio'n gweithio, ac mae'r gwych yn ôl iawn. Rydw i'n gweithio, rydw i'n gweithio ar ffínol. Mae'n credu'n ddweud yn gweld i'r gwasanaeth yma, yn ymweld i'r ddweud yn ei wneud, arno'r ddaf I wedi'u gwelwch, ond yw'r ffordd agorffent, a bod yn cael ei ddweud'r ffordd agorffent i'w ddweud yw'r ddweud yn yn ymdweud yw'r ffordd? Mae'r ddweud yn ddweud. Yn ystod, yr oedd oedd oedd oedd o'r ddweud yn digwydd o ddweud i'r ddweud, i'r ddweud i'r ddweud i'r ddweud, And pretty much everyone has something that they would like to protect. So I've had this conversation. I've had this conversation personally quite a few times with friends in family, usually after a beer or two. And it comes up, why should I care? I've got nothing to hide. So I usually just ask them, well give me your mobile phone and let me have a look through your text messages o'r ddatblygu sy'n hy wrth amddangos mewn gwirionedd iawn i ymddangos. Mae'n ddweud o'r tuogio eu gwirionedd iawn i chi, ac yna gyrraeth yn ei fod yn ei ddarpar. Felly, yn ymdill wedi fod yn y dryf yn oed i gyd. Felly mae'r ddaeth yn cull oedd o'r ddweud, a chydig i fynd yn dechrau'r ddechrau, byddwn ni'n ddif Weiniddo'r cyfrinwyr. Mae'n bod yn ddod, mae'n oed i'r ddechrau'r ddechrau'r gwyllt, i'w ddyn nhw'n gallu gallu'n creu andio gwahanol yn ddegwyd â'r mwygol o yma siarad a myngen amgylch. Mae'r ddau gwneud、 Gwyddoedd agorodd yn ddegwyd pan hynny. 1. Rhywb erbyn amlwg ein bod yn ddechreuwch yn ddegwyd, yn i ddegwyd o'r ddegwyd gan fairfyrdd mewn dda o'r ddegwyd ac o'r theid arall o'r ddegwyd? Felly, yna'r ddegwyd ar gyfer ysgrifennid, dod yw yn ddegwyd yn ddegwyd yma ym mwy ffordd gyda'r rhai, ac yn ymgyrch, y cyfnodol yn ymgyrch yn ymgyrch. A dyfodol yw'r rhagor. Mae'r cyfnodol yn gweithio ymlaenau, oedd cyfnodol yn gyfnodol, ac yn gyfnodol, oherwydd, y cyfnodol, oedd ymgyrch yn cyfnodol, a'r cyfnodol yn cyfnodol. Yn ymgyrch yn cyfnodol, mae'n gweithio'n cyfnodol. Felly mae'n gweithio'n cyfnodol yn ymgyrch yn dymwy'r cyfnodol, ac y dyfodol yn gyfnodol yw uwch yn ei ddawillon. Felly, oherwydd, bod gennych cyfnodol yn cyfnodol sy'n blaen gyda siaradau yn cael ei ddoeith hwnau. Felly'r ddweud o allan oherwydd yma sy'n gweithio i'r cyfnodol yn cael ei ddweud o gweithio sy'n ddweud o gweithio. Mae'r ddweudidion yw ddaach yn rhoi hyn, sy'n beth o'r добавd yn cael ei ddweud o hwyl yn cynnig yn gweithio. to get fixed. We also got manipulation through ads. A lot of people think, ads don't bother me if you're not blocking ads, ads don't bother me, I'm not influenced by ads, but study after study shows that people really are influenced by ads. And products is one thing, but there's other things that people show you ads for as well, which are somewhat more pernicious. Yn y gallu eich hwbeth, o hynod, o'ch cyfrifio'r ffordd, yn dweud gyda hynny, y gallwn, ydym ni'n digwydd ramdill. Byddwch wedi bod nhw'n gwneud y bwrdd ymarfer, oherwydd ar y reben o hyd o ffraith, a greu arwain o'r llefynol. A llai i ysgol, oherwydd, eich parnaf ni yn ymgarfod, mae'r parnaf ni'n byw iddyn nhw'n ceisio fel pobl amser hynny. … ac rydyn ni oherwydd iechyd ar gyfer teulu ei bod yn ymdrool. Ie nes deallr o ffiltr yn ogynt. Mae'r lleolau hyn yn eich bod ymwybod. A ffiltr yn ymwybod y ffiltro sydd yn cael ei fath o'r berthysgu rai cisiwys. Rydyn ni'n gweld ei wneud yw'r cunning hefyd Dream. Felly yn ymweld eu rhai ei fod yn ymweld ei gwaelwad, Cardiol yn ymweld'n ddyliadau erion hwn. A'r ddiweddau hynny yw mae'n cyfegau o'r rhai cyfnodau. can create or exacerbate polarisation in society. Additionally what many people don't realise is that it doesn't take a lot of data to be able to identify you. So an analysis conducted by MIT showed that it only takes four pieces of date and location data to be able to identify 90% of individuals from a dataset of 1.1 million credit card transactions over a three month period. I've been practising saying that. So it's critical to remember that privacy isn't just about protecting your single and seemingly insignificant piece of information which is what people usually say when they say I have nothing to hide. So for instance someone might say well I don't care if a company knows my email address or I don't care if a company knows where I am in the world when I'm buying something online. But when you're using these services it's not just that website that's looking at you necessarily as we know from trackers that are on the web. So that data is all being aggregated by advertising platforms like Google and Facebook. What they then do is form a more complete picture of you, what you're doing, who you're spending time with, where you are and things like this. This leads to data profiles which lead then to serious and significant privacy harms. The whole thing is a bit creepy. I'm sure you've had adverts following you around the web and yet it's creepy stuff. So we can't stress enough that your privacy shouldn't be taken for granted. I have nothing to hide response, it does that. It makes the assertion that governments and corporations should be able to access your data by default. But privacy should be the default. And at DuckDuckGo we're setting this new standard of trust online and believe that getting your privacy online should be as easy as closing the blinds. But it's not all doom and gloom. The good news is privacy has gone mainstream. Last time we checked, even though last time we checked American adults, majority of them didn't realise that Facebook owned WhatsApp and Instagram. I'm pretty sure everyone here probably does. But what we did find was that 24% of American adults now take actions to protect their own privacy online. And that figure is going up year on year, so that's a positive trend. In 2018 in particular was a good year for privacy. And look where we are today. We're here at Fosden on a dedicated online privacy track, which is extremely welcome. So how does open source help DuckDuckGo and our vision? Well, our company values are to build trust, question assumptions and validate direction. And open source allows us to build trust with the community through transparency and questioning assumptions and validating direction using community collaboration. We're also a small team, so there's only about 60 of us in the company. And so as a result we don't have a vast engineering capacity that some of the other bigger companies do. So free open source software lets us what we like to say is stand on the shoulders of giants. Aside from that, many of the open source projects that we use and software that we use shares a vision of the internet with us, which is embracing openness, freedom and transparency. And at the end of the day this is the internet that everyone at DuckDuckGo likes to play in. Apart from open sourcing some of our own software, we are committed to open source and free open source software. And since 2008 we donated something in the region of $1.3 million to privacy related projects such as the open source technology improvement fund, TOR, electronic frontier foundation, GPG tools, open whisper systems and many others. You can see a big list of them on our about page. Last year alone we donated $500,000 to these projects and managed to accumulate that up to $650,000 through a crowd sourcing privacy challenge. So given all that stuff, let's talk about something more specific that we're doing with open source and specifically our privacy essentials product. We released this a year ago, almost a year ago to this week which is quite fortuitous for being here. Our privacy essentials product is made up of three parts and you can install it on pretty much any device like I said. So if you go to our website with Firefox or Chrome you can install our extension and we've got apps for iOS and Android, you can install free of course. In terms of Android you can install it from the Play Store of course, but also you can get it from fdroid and directly from our GitHub repository. The signed APKs on there is the exact same thing that we deploy into Google. So let's have a quick look at each of the three major parts and then we'll see how the whole thing hangs together. So the first thing is private search and like I said we've been doing this for over 10 years and the idea really came about because we simply just didn't want to be tracked online and having these pretty ads following this around. So what does it actually mean to be a private search? Well there's a few parts to this. The first part is when you go to .go it's like we've never seen you before because we don't collect or store any information about you and that's our privacy policy in a nutshell. With regards to that we even went so far as to patch and recompile our NGX web servers to completely remove IP logging. So we deploy that software and it's impossible for us to just change a bit of conflict to enable IP logging which means it's impossible for us to build any kind of session history for our users. And open source is the thing there that's empowered us to build our principles directly into the software that we're building and using. Next, when you do a search we're obviously returning these search results to you. We also like to leverage third party APIs so for instance if you type in something like a Bitcoin exchange rate what we'll do is we'll talk to Kryptonator API, pull down the data and show you what we call an instant answer. It's like a box at the top. When all that Kryptonator oversees though is a request from .go it never sees anything from our users. So that's another way that we're protecting people. And then when you click on a result we actively try to avoid search leakage. So most people here probably know that when you click on a link it's going to send a refer a header to the next site along and that contains the full domain and path. So what we do by default is change the refer a policy to origin. So the next site will still get .go but it won't have any of the path. So query string parameters and all that stuff gets stripped away. And if your browser doesn't support that well we detect that and we fall back to like a redirection approach. In terms of growth 2018 was a great year for us. Our daily private searches left 63%. So at the beginning of 2018 we were doing 19 million private searches per day and that was up to 31 million by the end of the year. And to put things in perspective since 2013 the surveillance revelations and snow and all that stuff that's a growth of 723%. So more and more people are taking action to do private searches. And 2019 I just wanted to get this in here because we just had our stats back from the last month and we blown away those records already with 36 million searches a day and we topped a billion searches for the month for the first time. Okay so the next part is what we call smarter encryption. So while HTTPS is becoming more common it's still not fully out there yet and every website is configured for it and some websites configure it incorrectly. The other thing is that when you go to a website click on a link some websites will redirect you to their HTTPS version and that's great but it's too late by then because whoever's watching that network communication traffic or whatever can already see the path that you try to access on that site. So that means you're susceptible to SSL stripping and man in the middle attacks where they detect this redirection and then they send you to their fake version instead so you can use bank account details and things like that. So given that security is so important for privacy the solution really is to not let that happen so we go straight to HTTPS whenever we can. That's not as straightforward as it sounds. I'll go into that in a little bit more detail in a second. So from the search page we do that and also if you're browsing the web using one of our apps as you browse the web not just looking at our searches we'll also actively upgrade your HTTPS connections proactively so that you don't even send the HTTPS version if we know the website supports it. So the goal is similar to the EFF's HTTPS everywhere project in that it seeks to upgrade your connections. The main difference here is the coverage that we've been able to achieve while we maintain it. Right now our list contains about 6 million unambiguous rules for upgrading HTTPS whereas HTTPS everywhere is roughly in the region of 160,000 rules and of course is the HSTS preload list which contains around 55,000 rules as well. Now just to be fair we acknowledge that HSTS uses the include sub-domain directive so that's a little bit harder for us to quantify and likewise EFF's HTTPS everywhere uses regular expressions that's also not easy for us to quantify either. But the end result is the same. These lists are manually maintained and can result in errors and they age very quickly as well. So the way that we've been comparing the coverage is by looking at the way people click on search results on our search results page and what we've found is that when we compare all these things HSTS only covers about 7% of the search results that people click on and HTTPS everywhere covers about 24% but our list that we've built covers about 77%. So we've achieved this by creating a tool which basically crawls websites and it goes from website to website looking for websites that can be upgraded but we don't just throw any website into the list there's a set of criteria that must be met before we add it to the list so we basically act like a user clicking on the website comparing the encrypted version with the unencrypted version doing image comparisons to make sure things aren't broken and we only add the sites to the list when we're confident that they can be upgraded without breaking the web. So the result of that is 6 million unambiguous rules that we ship to our apps and extensions as well as use on the search results page and as you can imagine that's not a small file so if you assume something like average domain name length of 10 characters then we're talking about a 60 megabyte file which isn't a great thing to keep pushing to our apps especially over a mobile data connection so this was a challenge that we solved by basically creating a... Okay, just, yeah. Thank you. Is that there? Yeah. So that's a problem we solved by basically creating a bloom filter implementation so the binary file that we ship is probably in the region of 2 or 3 megabytes rather than 60 or something like that so the extensions of course are all based on JavaScript and there's a fair number of bloom filter implementations out there for JavaScript. In terms of the mobile apps, we wanted to, a couple of things we wanted to be able to use the same code in both apps and what we found was it was pretty tricky to find a good bloom filter implementation that wasn't too heavyweight or didn't have too many dependencies so we ended up actually writing our own for that. All the code for our apps and extensions is in our GitHub repository so feel free to go and check that out and see how it all hangs together. So the third part of our essentials product is the tracker blocking, of course. So when you're using our mobile apps or extensions we're actively blocking trackers, third-party trackers on websites. It's possible for companies like Google and Facebook to track you across the web as you know and a research project out of Princeton University showed that Google is on something like 75% of all websites. The next worst offender if you like is Facebook and they're on 25% of all websites. So pretty much wherever you go if you've got any kind of connection to Google they know what you're doing, they know what kind of things you're looking at, they're building these profiles about you. Even if you don't have a Facebook account it's the same kind of thing, they're still building these profiles and trying to work out who your connections are and all that kind of stuff. So our tracker blocking is the antidote to that. We just try and kill that stuff. The way that works is it's not really rocket science but the trackers are loaded via JavaScript or XHR requests, iframe requests, image sources and so on. When we see a resource about to be loaded we check it against a list and if it's a tracker then we'll basically block it. This has some challenges of its own. If we just randomly blocked all trackers then we would be breaking the web all over the place. In particular a lot of websites use Google Analytics, Google Tag Manager and these kind of things and even though they can be used passively a lot of websites actually hook into them in such a way that if you block them the website will stop working basically. So what we do is we inject a stub for those JavaScript libraries which we call surrogates, I think that's a term used elsewhere as well, which is basically a no op version of that library. Same interface, API, same application interface and we implement all the same callbacks so the website thinks it's doing its tracking calls and then just is able to continue without being broken. The extensions and apps, we all have the same tracking logic. We have implemented them slightly differently. The extensions, the checkout code for that is great because it's a really good example of how to build browser extension for multiple browsers from the same source. But the apps have implemented them slightly differently so on Android we actually do the tracker blocking in the native layer in Kotlin and on iOS there are a couple of options to us but what we've done on iOS is inject some JavaScript into each page which does that tracker blocking logic so it's more like how the extensions work. In the future we will probably move to the content blocking extension format if you're familiar with that. So tying all that stuff together, all those three things can be brought together in what we call the privacy dashboard and the idea here is to bring privacy awareness to the mainstream. So when you're browsing the web with one of our apps it's blurred out here for aesthetic reasons but you can see the grade there next to the address and so as you're browsing you can see the impact of a website's tracking and all that kind of stuff on your browsing and you can tap into that to bring this privacy dashboard open and then what we have on the dashboard is a grade so I'll come back to that in a second. We have all the things that have happened on the page so have we upgraded the HTTPS connection for you? How many trackers have we detected and blocked? What are the privacy policy of that website and we evaluate that and then we take all those variables and we've spent a long time trying to work out an algorithm which will basically sum all that up into a grade from like A to F. So if you're on an A grade website you know that they've got a great privacy policy they're not tracking you and stuff like that there's very few actually. Sorry. So again all that stuff in our GitHub repository and our apps. The final thing I want to mention here is at the bottom what we also do is keep a track of where we're seeing tracker networks as you browse so in this example you've got Google which is 70%, Facebook 32% and I think it's Comcast at 18% and that's across all your browsing so your mileage will vary sometimes you'll see other tracker networks higher up and you can tap on that and drill down and see what all the tracker networks sort of prevalence is on all the websites you've been browsing. All that data is just stored on your device we never send it anywhere and you can type it out anytime you want to. So what's the future for DuckDuckGo? So everyone at DuckDuckGo is really excited about our vision and working on these privacy products. So a big part of what we do is constantly improving those products. Amongst that there's two interesting projects I think that are going on at the moment. Firstly, we've decided to build out our own tracker blocking list until now we've been pulling tracker blocking lists sourcing them from various places but because it's such a core component to us tracker blocking now we decided that we needed to take ownership of the tracker blocking list so we've started building the technology to do that for ourselves. What that gives us is transparency and also some level of automation. That leads to the second project which is this automated site breakage tool. So what we do is as we're building this content blocking list up is we're running it on and off against websites to see what effect our tracker blocking has on the website and we do things like see what resources are being loaded and side by side image comparison. If we detect this with broken a website it means we can flag up a tracker blocking rule for manual review and take some proactive action there to not break the web basically which is really important to us because if people are trying to browse and we're breaking the web for them it won't benefit from the privacy that we're going to give them because they'll just switch to some other browser. Of course we are always focused on improving our user experience and making it easy for people to install our products and get going quickly a bit fast. This is the team. I just want to slip this one in quickly. We're hiring. This is not a remote team. This is not all everyone. This is most people though. We have an all hands meeting every year. It's about in May where we all get together for a retreat and each functional team has their own meet-up as well. We're currently hiring mobile developers especially Android developers so if working for 100% remote privacy focused and super fun company appeals to you then check out our hiring page. I'm just going to wrap it up. I'm running a bit fast, sorry about that. Like I said, Doug Doggo had been providing private search for over 10 years and with the privacy essentials product people can take control of their online privacy from any device and of course all of that stuff is powered by free open source software. If you're interested in working with us go to.gov.com slash hiring to see what roles are available. If you want to try these things out for yourself go to our slash apps. If you go there with your Firefox or Chrome browser you can install it directly from the page. If you want to install the iOS app it's on the app store as usual and Android on the play store or ffdroid or our github. You can find out more about Doug Doggo at slash about. Spread privacy is our blog. That's a great resource for privacy tips and news seeing what's going on in this area. Doug Doggo are on github of course that's where we put all our source that we can share and you can follow us on Twitter and if you want to connect with me on Twitter I'm happy to connect with you as well. And that's it. So thanks very much for listening. I prefer not to take questions if you don't mind but I'm really happy to talk to anyone directly. I've also got some stickers so if you want a sticker please go and grab one. I'm going to go to the bar and have a beer if anyone wants to join me. I'll see you in there. Thanks very much.