Slashdot Mirror
Making A Better Browser History
jbtule writes "Students at the University of Illinois have released TrailBlazer, a new user interface to represent your web browsing history. It lays out the pages you visit in a simple 2D map with thumbnails and summaries. The project took 2nd place at the university's annual Engineering Open House and a three minute video is available that demonstrates TrailBlazer for those who don't have Mac OS X Panther. TrailBlazer is implemented with Apple's WebKit on a bare bones browser, but this interface would probably be more useful if it were added to a real browser. This is a much better history than chronological lists of web page titles or crazy cubes floating around a 3D space. Hopefully Safari or /insert favorite web browser/ will do something similar in the future."
This is a great idea - a visualisation of the underlying data in a form far easier to recognise than the data itself. Humans tend to react better to visual stimuli (think a map vs a series of co-ordinates, and try to work out which location is farther away from you). Kudos to the authors for the inspiration.
This new idea tells us where we are in a better, easier-to-use way, and we like that. It can tell us where we can go/have been, and tracks the paths between these nodes on our cyberspatial plane [grin, sounds a bit OTT, but..]. Perhaps a cyberspatial compass combined with a cyberspatial GPS system. CPS perhaps
It's also interesting to see that the 'cool idea' is something to aid the browsing experience, not to replace it. It seems we're happy with the idea of 'click here, go there', but want more intuitive or rememberable (is that a word?) cues for the journey itself...
Simon
Physicists get Hadrons!
Do people actually use their browser history for anything other than:
a) Checking up on shared computers' other users porn-browsing habits
b) Tracking the links they've visited in the past.
Personally, I have a 25 meg history file going back I'm-not-sure-how-far which I keep around just so that links I've visited are a different colour.
Am I the only one who doesn't want to be reminded of some of the sites he's seen? Like *cx?
tasks(723) drafts(105) languages(484) examples(29106)
I don't have a problem with Opera's current implementation. Their quick and easy one button (F4) sidebar let's me quickly search by string for title, and arranges it by reverse chronological order. This allows me to quickly type in something like "google" in the filter, and show me every google search I've done in the past let's say 2 months. From this I can usually pick up any trail that I've lost and find a page that I've visited before with ease.
-Christopher Wu
http://www.christopherwu.net/
This reminds me of the game Relate-a-zon that uses the Amazon's related items lists ("customers who bought this also bought...") to create a pathfinder game in which you have to use common sense and a bit of wild guessing to find your way from one product to another. In the end it visualizes your navigation in a url path graph.
OpenSource, scriptable, customizable ad infinitum integradete IRC for spatial use and finally a good reason besides games to have a fast graphics board
Videos and images available
If all text (tagged by URL) was dumped into one file per month and made searchable.
That way when I am trying to remember where I saw the instructions for the excell driving game shown on Slashdot earlier I would only have to search the text I have seen, not try and use google (too many hits) or search by thumbnails and page titles... useless since it was posted in a pretty much unrelated subject.
Beep beep.
People may find immense visual interaction to be more appealing at first, but ime it can get cumbersome very fast.
My history is just that -- history.
If I want to go to a page I was already at, I'll most likely know when I went to it and can easily find it. This contrasts with Expose which helps you visually organize files currently being used.
I can see this having it's benefits (when I really need to find a poorly titled page), but I highly doubt it will redefine any standards.
It's a nice tool, but what do you actually do with tons of browsing history? Most of the pages I visit have content that is changing frequently or content that I only need to check once. It's not very often that I need a particular page I visited days ago. And if that really happens Google is my friend.
spidergraph (it plugs into Mozilla)
The common way of working with a browser history function is manytimes a frusturating experience.
It's bad to the point of borderline broken. Hopefully there are no IP issues (in the property sense), and this may lead to improvements making usinging browser history less like pulling teeth.
-Pete
Soccer Goal Plans
This looks like a good idea for browsing your history. However, I usually find items of interst through two metods - I either search or I browse. This will help me in the latter. If this was combined with a free text search (maybe a client-side google) they'd have a heck of a tool.
Underholdning.info
This seems very interesting and useful... yet it only won second place. What form of earth-shattering ingenuity won first place? I can't find it anywhere.
Safari uses webkit which uses konqueror base
I can't remember the last time I used mine. Is this a solution in search of a problem or do people actually use their histories?
It looks beautiful. So bloody obvious! Amazing nobody has figured this out before. I'm reminded of something a former boss of mine used to say: "It took 80 years after the invention of the printing press for someone to figure out page numbers are a good iea."
Really, I could probably come up with a whole range of criticisms, but why? This is a great idea. Practical, obvious, useful. The most negative thing I can say about this is probably that I feel sorry for the inventors. They'll probably be forgotten after Microsoft and the Mozilla foundation have released their own unspeakably crude and complexified implementations.
Hopefully Safari or /insert favorite web browser/ will do something similar in the future.
And the organization who owns that browser will then patent the technology and own IP rights over it. It's an inevitability with the current business practices right now.
It would be nice if the results of Google could be filtered using your browser history.
This way you would have your own like WWW to search in and would only return sites you have visted in the past.
Net sa best, mar it koe minder
Webkit is the same data loading/display engine that powers safari. It's provided by Apple, works as far back as 10.2.6, and is part of the default 10.3 install. The XCode developer tools let you drag and drop webkits onto window prototypes - there's an article on doing so here:/ 01/23/w ebkit.html
http://www.macdevcenter.com/pub/a/mac/2004
So no, no Gecko.
JT
______ This mind intentionally left blank.
So if you wanted to search for an article on how to weave baskets (I know, weird example) that you had seen in the past, you could tell Google that you're interested in crafts and such but not shopping or historical research. Google would then filter out sites about the history of basket weaving and sites selling woven baskets. That doesn't leave a whole lot of room for anything but articles on how to weave baskets yourself.
I'm sure that'd get you there [almost] as fast as a Google search on your history. You'd probably come across even more relevant information than if you only searched your history. It's just another technique for Google to deliver the most relevant results in the search engine industry.
-- If you can read this, you are too close to my signature.
which is a Gecko base
The underlying HTML technology beneath Safari is KHTML, not Gecko.
-- Help Digitise the Public Domain at DP.
This paper reviews history mechanisms in web browsers back in '97. One of the mechanisms mentioned, MosaicG is stunningly similar to the work in this article.
MosaicG was released in 1995.
It's interesting though that Tauscher's paper (the first link) conlcuded back then that the 'stack based' histories we used were not optimal, mainly because sibling history branches disappear. She found that the best method tested was to have a 'context sensitive web subset', ie a graph showing the relationships between visited nodes in relation to the current node, rather than a strict history.
It would be really cool if TrailBlazer we able to integrate with Safari rather than act as a stand alone browser. If TrailBlazer was able to follow your trail by parsing Safari's cache it would be totally awesome. As it is now, TrailBlazer is a cool novelty, but as a browser it lacks many of the features most modern web users use.
I've been thinking about this for weeks ...
I realize that I still reflexively bookmark good sites, even though I almost never use the bookmarks (beyond the few I put on the Links/Personal toolbar). I just go to Google for everything.
Yet I'd like to harness the value of that information. I wish I could do a google search, limited to the bookmarked hosts. Weighted by how many bookmarks I have for that host. So if I have 30 bookmarks at 4guysfromrolla or whatever those results come to the top.
Should be possible by running a daily script on my bookmarks.html, to build a searching script. Do individual site searches for each host, ordered by the number of times the hosts occur in the bookmarks, then merge the results. Just haven't got around to it...
Not being a Mac user, I have a middle button {that isn't meant to be a flame, although it does sound like one ..... but bear with me and you'll see} which lets me follow links in a new tab. So I can keep a bit of a chain of thought together ..... if I know something is relevant to the page I'm reading, I can call it up and not lose the current page.
Typically, I'll do a search, then open one result at a time in a new tab; if the article is useful I'll keep its tab open, if it's no good I'll close it. However, it all gets very unwieldy once you have more than about half a dozen tabs on the go at once. Plus, tabs are {TTBOMK} not rearrangeable -- so the structure breaks a bit, because I can't put the tabs I opened from each first-level click next to one another. Tabs are good, maybe even great, but they aren't perfect.
Other times, I will bookmark a site which, on further exploration, turns out not to be any good. Which is a waste of a bookmark.
The computer already knows what sites I have visited, how long I spent looking at each one, whether or not I did any word searches {and what they turned up}, and what I clicked to next -- whether it was a link from that page, or if I returned to a previous tab, or started a new search. Now, if I want to find a page that I know I visited recently, how should the above-mentioned information be presented to me so that I can find the page I'm looking for, quickly?
It's good to see that question being addressed. This could be something the web has been waiting for.
Je fume. Tu fumes. Nous fûmes!
This sounds like a great feature, but like tabs (which is very helpful) - few people will use it to it's potential.
I think Apple needs to concentrate on being faster and more stable - I really even wish they would remove some features from Safari - when explaining ANY browser nowadays to my LEAP program classes (who are mostly elderly) - it is difficult, at best, to explain ALL the features - something you HAVE to do so they won't be confused and know where to go to set preferences, etc...
I also would like Apple to remember thet their core is only as good as the bushel - meaning - I would like Apple to realize that Opera, iCab, and Ominiweb give Apple the appearance of broader support - therefore, these should be features for those browsers to integrate - features people should want to pay for if they use them.
Yell & scream & rant & rave... it's no use... you need a shaaaave ~ Bugs Bunny
Seems awesome, i am even more proud that they did this on OSX first. Although, development with Xcode is extremely easy.
;).
Anyway, how are they going to deal with people that visit thousands of webpages a day/week. Is their history going to catalog all of those images and take screenshots of each one? It seems like there would be some huge memory/efficiency requirements and would make the browser more unstable as you visit more website.
Apple's own safari has a similar problem with web icons even, let alone whole thumbnail images!
Although, it would make searching through all my previous porn a lot more fun
-------
artlu.net
<ECODE>
I never understood why this hasn't been done before. But a lot of times you go backone or more pages, go to another page, then all your previous stuff is lost.
We need a way to say, from this page you want to these places.. Currently we are limited to:
S--->--->--->--->--->
But we need:
S--->--->--->--->---> +--->
+--->--->---> +--->
+--->--->---->--->--->
Where the '+' are junctions where two mor more links were followed
</ECODE>
Slashdot's rate-of-post filter: Preventing you from posting too many great ideas at once.
Two-dimensions are all well and good, but, c'mon, this is the 21st century. If I can't have my "Minority Report"-style glove-and-goggle mix, at least I can browse in 3D! This actually came out a couple of years ago. Fun to play around with.
Ryosen
One man's "Troll, +1" is another man's "Insightful, +1".
You're thinking of ProjectX, which was renamed HotSauce. If I remember correctly, it was a "fly through" of spheres in fake 2D, each sphere representing a site (and surrounded by smaller spheres representing linked sites). As you approached a satellite sphere, you would begin to see its links come into view....and so on, ad infinitum.
Here's one link I found in Google "apple hotsauce browser."
this winds up being too large to show any useful information in a reasonable area (i.e. the size of a screen). I would have thought that expose would be a better solution to this problem...
With all of the privacy baggage aside, imagine providing a browser history to your counselor to see how effective a particular treatment might be.
-- In Soviet Russia, radio listens to YOU!
This is a great idea - the only problem I can see with it is that the browser history map becomes too unwieldly, requiring a lot of horizontal and vertical scrolling. The missing element is mouseover magnification like the OS X Dock has - that would let the user see their entire history (OK, let's be realistic - one week at a time) in the window, and then home in on the relevant part by moving the mouse. Kudos guys!
I don't know if this is a personal quirk of mine or if other people share similar cognitive function, but I remember colours better than anything else. No matter what else I remember about a webpage, I always remember its background colour. Therefore the thumbnails would be invaluable to me when I'm trying to find that page I saw with a white background and a big pic that was pinkish-white (Caucasian skintone) with a red dot in the middle.
This sounds very much like the "web map" feature of IBM's "Web Explorer" browser for OS/2 back in the early 90's. Web Map created an html page comprised of the links you had visited in that session, arranged in a heirarchical manner. I've not seen anything quite like it in other browsers.
Cool. Now we need a mozilla plugin to do just that. Any volunteers?
Now where's the copycat mozilla plugin...?
A hierarchical (and usually enormous) tree of bookmarks is a broken, broken, broken concept. I spend more time searching a bookmark I know I have, that looking for it in Google. That means something: Google is a better tool than bookmarks.
What I'd like to have is a powerful, a-la-Google context search of my history: I don't want to save "bookmarks", I want to drag predefined "keywords" onto TrailBlazer's history thumbnails; so that when I later select a keyword, all pages that I've marked are retrieved in their full browsing context.
Singularity: a belief in the "God" idea with the "demiurge" relation inverted.
So bloody obvious! Amazing nobody has figured this out before.
FYI - this feature (more or less) was on IBM's initial web browser for OS/2, before they ported Netscape. I remember using it almost a decade ago. I think they called it a "Web Map" and it was just a tree representation of the pages you had opened. Very simple, intuitive and useful.
Suppose you've written something. Then, you undid the last sentence or two. You wrote something else. By doing so, you've essentially deleted the "redo" information of the sentences you originally undid. Therefore, if you don't remember what you had written the first time, there is no way to "undo" this "branch" and go back to the other "branch" unless you had originally planned to do so and saved the file, or copied the original text into a buffer, or something along those lines.
I imagined that the undo/redo information would have to be structured somewhat as a tree, or hierarchy, of edits, much like CVS is structured with multiple branches and the ability to fork, merge, etc. To solve all the problems that I foresaw, the model became pretty darn complicated (about as complicated as re-implementing all of CVS inside of the undo/redo feature, plus supplying an interactive user interface for this mess), so I never implemented it.
I suppose that at some later point, when I began browsing the web, I thought that something similar should exist for browsers. Every time you go "back" and go on a different path, you basically create a branch. But eventually, I came to the conclusion that having just the simple "back" and "forward" feature has some advantages over a branch-enabled navigation feature. For one, it is much easier for non-tech-savvy users to understand (if they even know the feature exists--many people are really only aware of the "back" button). And furthermore, it allows you to cover up your tracks, to some extent. Say, you're at a public library, and you just used their web browser to find something. And suppose you can't delete the internet cache because of security settings on that machine. So you go "back" a few times, type in a new web address, and you've essentially erased your tracks, as far as any patron without administrator access will ever know. (I assume that if you can't erase the cache, another non-admin patron can't read it.)
I would still love to have this feature in my web browsers (and text editors), as I like to have lots of windows open all over the screen, and I juggle from one to another, and routinely go back and forth many times. With the ability to go back and open another "branch" in another window with a few keys, I think I could be a lot happier with my web browser.
In other words, if a slick interactive way to do this kind of stuff can be implemented, then I see at least two applications (browsers and undo) that would greatly benefit from such a thing.