loading . . . Reading This Is For Everyone by Tim Berners-Lee I have started reading a book by Tim Berners-Lee (or TimBL as hid early internet nickname goes). I’ll be posting some interesting quotes here.
## Tim-Berners Lee on AI and what it can bring
> The AI wave is one of our biggest opportunities yet, and promises to deliver a great deal of value for humankind. But experience suggest we also have to be careful; this technology is so powerful that it threatens dystopian outcomes.
## Tim-Berners Lee on open web
> Over the past few decades, I’ve fought to keep the web transparent, open-source and freely accessible. […] the web, a small, but significant part of it – the addictive forms of social media – have multiplied into something misleading, toxic and habit-forming. That’s pretty far from my vision. But because this small part of the web is so addictive, people spend a lot of time on it and as a result most web traffic is now concentrated in a handful of large platforms which harvest your private data and share it with commercial brokers or even repressive governments. That’s pretty far from my vision too.
## Tim-Berners Lee on usability and related universality of the web
> Watching the resistance to CERNDoc was an important lesson for me. If a documentation and storage system was to be acceptable to all the folks in that room, then it would have to make no demands of its users. It would have to impose no constraints. It would have to work with any format, any computer, any operating system, any categorization system and any sort of document. It would have to be a sort of universal space, in which all of their different systems could coexist. That universality would later be key to the web.
## On the name of World Wide Web
> I decided my project needed a catchy acronym. Some names I considered, but rejected, included the Mine of Information (MOI) or The Information Mine (TIM). I ultimately decided to call my system the ‘World Wide Web’.
>
> Many people at CERN grumbled at this name and complained about the nine-syllable ‘WWW’ acronym even more. Who makes an acronym — a shortening — that takes longer to say than the original name?
## On how HTTP protocol works
> Later, these two technologies – DNS and ΗΤΤΡ -would become conflated, but I think it’s important to be clear about what I invented (URLs, HTTP and HTML) and what I did not (the dot-com names). What I did do was to harness the power of the DNS approach, so that if the (now-defunct) Symbolics Corporation of Massachusetts wanted to host a web page, it could do so with the name
>
> http://www.symbolics.com/sparkle.html.
>
> When you click on this link, your computer looks at the URL, saves any ‘#section1’ bit for later, and takes up the first bit, ‘http:’. This tells it to prepare to initiate a connection to another computer, using the hypertext protocol. Then it takes the next bit and opens an internet connection to the other computer, whose name is www.symbolics.com. Then it sends the rest of the URL, sparkle. html, to that computer, and that computer delivers back the contents of the web page – going to Section 1 if there was that bit after the hash sign.
>
> ## On domain addressing system
> A number of decisions I made in that first year of web design had far-reaching consequences I didn’t intend. First, by relying on the domain name system I ended up tying the web to the instability of domains.
>
> There was an alternative to DNS which the UK had been using, where the parts were largest first, like ‘com.microsoft.www’ rather than ‘www.microsoft.com’ – but that was not what the world was adopting. So I chose the American way, although ultimately the British way would have made more sense.
>
> Here’s what I mean: logically, web addresses should become more specific as you go on, so it should really read ch.cern.info, since ch’ – the country code for Switzerland – is more general than CERN, the research facility in Switzerland. Under this scheme, a business address might logically read, uk.co.widgets.www
## On double slashes in HTTP
> Lastly, using the rather obscure Apollo naming system meant placing the two slash characters ‘//’ between the http protocol specification and the DNS address. The idea was that if you were making a link between two websites, then you could skip the ‘http:’ and just start with ‘//’ – but in practice people don’t. So I could have designed the web without the double slash, and that would have led, over three decades, to quite a lot of saved keystrokes. Sorry!
>
> Then again, maybe early adopters, people who had seen other systems with double slashes in them, would not have picked it up, and we would not be using the web today. Who can say?
## On hypertext language – HTML
> This is where the third part of the design of the web comes in, as the web page has to be encoded in some language. At the time, there was no standard format for hypertext, so I defined one: the Hypertext Markup Language, or HTML. Again, the rule for designing the web was to use existing stuff wherever possible. Much of the original version of HTML was based on Standard Generalized Markup Language, or SGML…
## Javascript could have been Python
> I began pressing developers to come up with competing browsers, ones less in thrall to shareholders. In 1995, Guido van Rossum introduced Grail, a browser that would let you run applications within itself using his Python language. I met with Guido and encouraged him to develop Grail into a real contender – but, in exchange, he wanted me to start pushing Python inside W3C. I didn’t feel I could justify endorsing a programming language which, at that time, fewer than 1 per cent of software developers had ever used. A mistake, as it turns out – today, Python is the most popular programming language in the world.
>
> The good news is that, through W3C, we were able to bring both these sides to the table, preventing a blue text/green text standards war from erupting.
## Global collaboration
> The connectivity of the web accomplished something else I really hoped for – it allowed people to make connections across barriers of class, ethnicity and culture. (Language was a harder barrier to cross, although I’m optimistic about developments in instantaneous machine translation.)
>
> My vision for the web was always that there might be someone in, say, Accra, with half a solution to a problem in their head, and someone in, say, Vancouver, with the other half. I would call this group collaboration, this inter-creativity, the web’s core principle. Perhaps, with a little compassion, it may yet lead to the resolution of global political problems.
## XML vs RDF
> This divide had a big effect on the web and the industry. XML data looked like a tree you know, static, hierarchical, with everything in a folder. RDF was more like the web, a dynamic graph, where connections might be drawn between any two pieces of data, even unexpected ones. The second approach is so much better. One of the reasons that Google search results are so smart about things like airline flights and facts about people (height, age, title) is that they have built a huge graph of knowledge, accumulated from all the data on the web.
>
> Graph versus tree: you would not believe how intense the debate over these competing approaches was.
## Public data and open data
> Even as some kinds of private data that shouldn’t have been shared were shared, other forms of data that should have been shared weren’t. Every year governments collect reams of data in maps, bus timetables, government spending breakdowns, records of traffic accidents, environmental measures and economic statistics. This data belongs to citizens, as it is paid for with taxpayers’ money. With the exception of sensitive personal data and matters of national security, most of that data deserves to be shared with the public and academic researchers. Governments are great at gathering information, but they are generally pretty terrible at analyzing it.
## On choosing sides
> Give me a choice, and I will take the side of Robin Hood over the Sheriff of Nottingham.
## Collaborative filtering
> The ability of collaborative filtering to find you something you like is great. Spotify, following the lead of Last.fm, uses it to automatically generate playlists and radio stations customized to user preferences. Amazon and other online retailers incorporate the collaborative filtering technique into their product recommendation services, greatly improving the online shopping experience. Instagram, Pinterest and other social media sites use collaborative filtering to provide personalized recommendations.
>
> What I was less excited about was the ability of collaborative filtering to find what agitates you. I saw this happen on platforms like Twitter and Facebook, which were optimizing for _engagement_ , rather than enjoyment. It turns out that if you feed people provocative content, you can keep them on your platform for longer. So the AIs started showing nastier and nastier stuff – ‘rage bait’, basically – which made users angry and argumentative, but kept them logging on.
>
> Over time, this collaborative filtering contributed to deep political polarization.
## Social media and addiction
> Social media is a fantastic innovation with tremendous potential to bring people together. And given the vast amount of information uploaded to social media every day, we need algorithmic agents to organize the media that we see. All we need is to regulate the _addictive_ algorithms, the ones that guide people into rabbit holes. It’s the technical design decisions of the social network sites that drive this kind of polarization that we’re targeting – it’s the algorithm that’s been trained to produce this outcome. We need to change that, one way or another.
## Interoperability
> For instance, email apps all work together – interoperate – because they all use the same email protocol, which is an open standard. Podcasts and calendars interoperate too. Social networks don’t. Why can’t I share my Facebook photos with my LinkedIn colleagues? Because Facebook, LinkedIn, Instagram, X, are all silos with no interoperability. Accordingly, we need a new means of connecting and powering the social media we use. We need new open protocols and apps — a pro-democratic layer that empowers users and combats disinformation. Not only that, but to attract users, the new layer has to be enduring source of value. The result could be a resurgence of optimism and faith in longstanding institutions.
## Data layer (for the web)
> Although we have HTML for documents, and URLs for making links, we have never settled on a standard for the _data_ layer of the web. […] The Solid protocol (solidproject.org) we’d been working on addresses these issues. https://ambience.sk/reading-this-is-for-everyone-by-tim-berners-lee/
