It's a good question, @masukomi . I think the conceptual friction comes when you are trying to create websites on dat. Websites were built for the web, and naturally thrive. Instead, one should make a dat site. In other words, the intention and design changes and what that change allows for I find to be really exciting.
What I mean:
If you like a dat, you can seed it. Now it's available from at least two people, increasing the chance that it'll be available for someone else who seeds it too. The seeding option comes up automatically when you bookmark an archive in beaker, so if you like something enough to want to revisit it, then you are encouraged to actively show that appreciation through this infrastructure support. This changes the creator/consumer model in a real nice way.
If no one is seeding your archive, and your computer is not up, then it is not available. That's okay! Ephemera is beautiful. Live shows disappear once the band stops playing, zines disappear from print when the last one is bought and the zinester chooses to not make another printing. Live shows and zines still resonate with people long after they're gone. Permanence is not a fundamental quality of life, and it doesn't have to be a fundamental quality of dat design.
The archive also doesn't disappear as it lives now on your computer, either in a writeable form (if you're the author) or read-only form (if you're a seeder). On top of this, it is supremely easy to not just put up a dat site, but to put up a thousand dat sites. You don't have to centralize your own work. You can create a new dat site for each new need or idea, and let each one spread out as it will.
It's also supremely easy to fork a site, which puts all the files on yr computer but now writeable, which means much of the contemporary dat design is for building sites that double as tools. Dropout, by Jon-Kyle, is a great example of this( dat://dropout.jon-kyle.com/). You can visit his dat site of articles he finds interesting, fork it, and now have a tool for making readable, offline versions of the sites you find interesting. tiddlywiki on dat is another incredible example--having an offline wiki that has all the wonderful tools of its online twin has been huge for me. The dat api is really clean, and makes it easy to read from any archive so you see a design pattern where you have a set of personal tools that you use to interface with another person's small set of tools, along with your more ephemeral one time products.
All this means means dats become closer to songs or stories than webpages, where the act of sharing it is not just sharing the thing you have to say, but the means in which you said it, empowering the visitor to share their own story inspired by yours.
And for discovery of sites, the answer is Scuttlebutt! For me, Scuttlebutt is integral to Dat in a couple ways. If I make a new dat zine, I will immediately post it here. Now other people can read it, which I love, but I also have a record in an immutable log being sustained by all of my friends that says: "On This day, at this time, I said that I was the author of this cypherlink." If someone were to fork my zine, remove all references to me, then try to sell it as their own in a way I did not jive with, I would have tangible proof that the contents of it were mine, and that I stated this far before the copy's appearance. I am not sure of the full legal side of this, cos it's an avenue I don't care about that much, but it's nice to me that content authorship is embedded within the code, and your proof of authorship is supported by your butts community.
All in all, I think to see dat's potential you should not look at the web, but at cultures that flourished before the web, and how they could re-emerge now. Supportive, independent, regional* culture dies on the web today but could thrive on dat and beaker.
- My use of regional does not mean geographical regions, but a type of cyber-geography that is broader than just 'shared interests' but still has a unique set of values and histories that unites the people within it. This scuttlepelago is regional to me.