On Tuesday 09 September 2008 00:49, Colin Davis wrote:
> 
> I worry that right now, Freenet doesn't provide a very good first user 
> experience- I worry that this leads to frequent installation of the tool 
> out of curiosity, and then subsequent uninstallation when the user can't 
> find anything fun to do with Freenet.

Agreed.
> 
> The main problem is that Freenet is marketed, and fproxy is oriented, as 
> a network layer tool. This makes sense, to a degree, since the goal is 
> to have many applications depending on freenet, but in practice users 
> download freenet itself, find it lacking in wizbang features, then 
> remove it. I propose that it may be in the interests of the project to 
> tweak fproxy to cater to that initial installation experience. There are 
> a few tweaks that can be done that would dramatically improve user 
> experience.

Strongly agreed!
> 
> 1)
> For instance- Surveys have shown that people install freenet but 
> unisntall it because they can't find a way to find "files" to download.. 
> I think the right solution to this is NOT to bundle Frost/etc, but to 
> make it
>  drop-dead-easy to publish a flog on Freenet, and include whatever files 
> or collections of files you want in your flog.. 

Unfortunately, we don't have an easy to use web interfaced flog engine. I'm 
assuming it would have to be web interfaced. Thingamablog has an external UI. 
Echo is way behind where it needs to be to be a usable flog engine. And once 
we have a basic flog engine, then we'd need to add support for attaching 
files to it. However, this is an attractive idea, especially if we can tie it 
to the web of trust...

> By doing so, and  
> providing links to them, you can do two things-
>             a) There is an incentive to the publisher to continue to put 
> out new editions of their site, since they are the "goto" place for 
> certain types of files
>             b) There is a single-version of the file that is popular, so 
> it becomes better propagated.

c) More content, hopefully long-lived, on fproxy.
d) Can have reviews etc as well.
> 
> Imagine for a moment that I want to start a new page Nine inch Nails 
> remixes[1].. As the provider of this page, I can create links as I want, 
> rate them, provide reviews, etc.. This adds substantially more meta 
> information that files in a Frost store.

Indeed. But this means somebody has to write a good flog engine, including a 
good user interface for adding files. Do current web based blog engines 
support this kind of thing? Do they do it well? We may be able to take code, 
templates or at least ideas from deployed systems...
> 
> Further, as new remixes come out, I have a great incentive to add them 
> to my page, since I know that people around the world are coming to my 
> flog to listen to them.

Right, provided that the search system efficiently finds the page.
> 
> 2) Once people can easily publish pages and add content, we should try 
> to make it easy to find the files that people are already adding. The 
> default bookmarks are a good step, but the Librarian helps a lot more. 
> As I mentioned above, people love Search.. We can give it to them, in a 
> way that doesn't violate the Freenet model.
> 
> If we have the Librarian-style search on the default fproxy page 
> (perhaps only appearing after it has grabbed a few indexes), we can give 
> people a go-to place to look for information.

A potentially serious problem with using XMLLibrarian for everything is that 
if it starts to be attacked, it could quickly become unusable. Maybe we can 
tie it into the Web of Trust... Also XMLLibrarian does violate the Freenet 
model - it's like an index site, it has to be published by a user who spiders 
the network.
> 
> We can further increase the usefulness of this Librarian by 
> automatically checking the flogs and freesites we go to for a published 
> spider file. If Flog authors started including their own XMLSpider files 
> in their flog, urls would propagate quickly. We could add a checkbox to 
> the publish tool that says "[ ] Help others explore Freenet by 
> publishing a list of all known sites"

Doesn't that imply they run XMLSpider? Generally most users won't want to run 
the spider... If it's just based on their browsing there are serious privacy 
issues.
> 
> 3) There NEEDS, NEEDS, NEEDS to be a reliable way to exchange data.. If 
> it's Frost-style messages, FMS, Freemail, Whatnot, this should be built 
> front and center into Freenet, and placed right next to the Create a 
> Freesite link.. Having a reliable way of sending data, not just to your 
> friends, but to an anonymous freesite, is crucial for interaction.. 
> Essentially, we have weblogs working, but we need comments..
> 
> The reason for this is that we need a community to build up around 
> specific flogs. We want people to check in with a flog every morning, 
> read the links and news, and leave comments and suggestions for future 
> sites. There needs to be one solid way of doing contacts for this to 
> work at all ;(
> 
Okay when you say data here you mean chat. Yes I strongly agree that we need a 
good web based chat system integrated into Fproxy and embeddable in 
freesites. Unfortunately no such system exists right now!

> People need motivation to keep posting.. Hearing from users helps. Lots!
> 
> 4) Minor issues-
> 
>        a) Can we Queue to Downloads, or Downloads and Uploads? I know 
> what a download queue is, but even I wouldn't necessarily look for it here..

Agreed. We have so far used single word descriptions (in the english version) 
to save space, and assumed the users will be inquisitive hackers and try 
everything, or at least see the tooltips. But even if that assumption is 
correct, they will find what they are looking for more quickly if we have 
adequate descriptions.
> 
>        b) Why are Statistics and Connectivity given the same high-level 
> placement as "Configuration" and "Downloads"?
>              These are fun, and geeky, but they aren't interesting to 
> most users.. The top-left placement is very high-value real-estate.. 
> This should be used for links like:
>               "Publish a Flog!"

Doesn't exist.

>               "Search Freenet!"

Largely doesn't exist.

>               "Private Message your friends"

Does exist but clunky. Do you think it is worth putting some effort in to 
this? Messaging your Friends and messaging your anonymous friends are two 
rather different things.

Some thoughts on UI:
Provided that opennet works, and puts up alerts if there is a problem, there 
is no reason to have Strangers at the top level. Statistics should be in some 
advanced section with Strangers and Connectivity. It's an open question as to 
whether Configuration should join them. Configuration is certainly going to 
be used by the average user, mostly to change the security levels or the 
bandwidth limit. Alerts doesn't need to be top level either as the user can 
click on the alerts for more details (and we do try to make this obvious 
now).

So:

Browse Freenet -> Freesites page, with the bookmarks and a prominent search 
box.
Chat -> FMS web interface
Mail -> Freemail web interface
Filesharing -> Searchable WoT-based filesharing system
My Friends -> Friends page but cleaned up in simple mode, with obvious and 
easy to use messaging, and self-documenting.
My Freenet Blog -> Easy to use blogging page
My Downloads and Uploads -> queue page
Settings and Status -> sub-menu:
        Configure security levels
        Configure other settings
        Status alerts
        Statistics
        Plugins (better name???)
        Internet connectivity
        Connections to strangers -> if opennet enabled

Or do you think we should include the config options on the main menu, and 
have the submenu be strictly geeky stuff?

The catch is that if we strip out what we haven't already implemented, it's 
going to look rather sparse. Maybe that's a good thing, with appropriate 
theming?

Browse Freenet
My Friends
My Downloads and Uploads
Freenet Settings and Status

>        c) Freenet should model it's UI  less after router configuration 
> pages, and more after LiveJournal/Wordpress/Etc.. They know how to make 
> publishing sites look easy and fun.

I was thinking more social networking sites. But in any case we should improve 
the GUI.

>        d) There should probably be some way to allow a node to pre-fetch 
> as much as possible..
>              I have over a TB of storage that I'm willing to let Freenet 
> have.. So why is it still slow when I browse requests?

Lots of reasons. There is a prefetch option implemented for fproxy, but when I 
have tested it it has not improved first-time-user performance. There are 
however a number of ways to improve overall network performance; we have 
implemented FOAF, soon we will implement no-swap-on-opennet, combined we 
should be looking at a reasonable factor in terms of performance and data 
retention/reachability.

>               It would seem that if I'm integrated into the routing, and 
> I start seeing a lot of requests passing past/through me, I could start 
> grabbing the pages referenced by those keys, just in case I need them.

No, you can't, because you can't decrypt the requests.
>       
>     e) Opennet is referred to differnetly in different places- Strangers 
> on the main menu, Turning it on is promiscous mode, Insecure mode, etc.   

Agreed, this is a problem. However, users don't really need to worry about 
opennet any more, now that we have the security levels system. I have just 
made opennet an advanced option.
> 
>     f)  The Plugins page is too geeky.. Ideally, this would be more like 
> Firefox extensions. We don't need the FQDN of the extension, etc.

Agreed, we should make the plugins page simpler (and maybe have an advanced 
mode). We should also include a brief description of each plugin that has 
been loaded, and each official plugin that could be loaded.
> 
>     g)  Freemail has no instructions on how to use it.. None of the 
> plugins do, really.. No one is going to figure these out...  ;(

Freemail in its current state isn't usable by most users. They expect webmail, 
and that's what we've got to give them. Dave Baker's student made a start on 
this, but sadly didn't finish. :(
> 
>     h) Further, to be a consumer-level tool, it'd be really great to 
> unify identity files a lot more.. Your flog id, is different than your 
> Darknet ID, is different from your FMS id, is different than, is 
> different than, etc.
>         It'd be best if we could have one public/private key, and just 
> have fproxy use it for lots of things behind the scenes, so we can 
> publish a blog under the same username we use to connect to our darknet 
> friends, etc.

I agree that sharing privkeys between many apps is a good thing. But IMHO the 
main way this should happen is through different applications using the Web 
of Trust, although we should consider reusing keys for flog's.

>         Then, we can have a "Freenet ID", and be proud of it, whereas 
> now, again, we come across as a platform, not a destination.

Your Freenet identity for connecting to Friends is traceable, so cannot be 
related to your Freenet identity for anonymous chat etc.

Another point about user interfaces: What about different identities? Many 
users will need this. Do we want to have a global identity selection, which 
determines who you post as etc, visible on all generated pages, maybe with a 
different theme for each one? We could of course disable this if the user 
doesn't use multiple identities.

If WoT identities are integrated to this level, we should probably create the 
identity in the first-time wizard. Perhaps plugins can hook in here too, and 
if loaded later on they can show a setup page when they are first loaded?
> 
>      i)  Is up&p turned on by default? We should enable it, if they 
> choose low security...

UP&P is on by default. Unfortunately it doesn't always work.
> 
> [1] Reznor has released certain tracks under a CC license.

Attachment: pgphEXkWSIUBh.pgp
Description: PGP signature

_______________________________________________
Devl mailing list
Devl@freenetproject.org
http://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to