Browser Sedimentation | Hacker News Reader

Browser Sedimentation

tbray.org

73 pointswglb42 comments

A wonderful example of this is people who type URLs into Google Search or search queries into address bar (before most browsers had Omnibar-like features). To all of us here, it clearly makes sense that the browser is a local software that enables you to visit websites while a search engine is a website that finds other websites. Typing a full URL in Google is generally a waste of bandwidth and time.

This abstraction of service is lost on many users who see the browser and search engine as the same thing. And even different websites and the Internet as the same thing. I've had to field many tech-support calls because a Java exception occurred on the outsourced Payroll processing site but somehow since I manage the IT Department, clearly it is something I could fix.

I think a good way to look at this is to pick something complex that you use regularly but don't know much about, like an automobile or cellphone hardware. Why does my cellphone suddenly lose reception for 5 seconds when walking under my carport? I have no clue but I suspect it has something to do with the signal or maybe because there's a lot of metal around. That is how users feel when you show them a big red X, regardless of how much descriptive text you present. Best is to avoid situations where you have to show X and just make it easy to revert to the previous situation.

While I largely agree with your comment, I think the story of you and your cell phone differs from that of the user who calls you asking to fix the website, in that you know that you do not know what is causing the signal to degrade.

I have nearly infinite patience and can completely understand someone coming to me with the following:

"hey, I have no clue what just happened, and you probably aren't even the right person to ask this question anyway, but I figured that you might at least be able to give me a better understanding of what went wrong if not: I tried to access my payroll today, and it didn't work; is that something you can fix?"

That is the kind of message I have myself sent on occasion when I'm "in over my head" in a field I don't understand, and have run into issues that block progress.

However, the kind of message I get on a daily basis is:

"you are a piece of work: I should report you and your organization to the police. how dare you have given me this error message! I demand that you fix this situation in the next 24 hours or I am going to go on Twitter and Facebook and make it clear to everyone what kind of crook you are."

(Meanwhile, they are complaining to me that... their bank account is out of money... and the error message came from PayPal. I even get users complaining to me that my software took a $35 fee from their account because of their lack of funds, which is obviously an overdraft fee from their greedy bank, and has nothing to do with any of the software in the stack that is billing them.)

This person believes that they know the things that they in fact don't know (or possibly isn't even able to contemplate that there might be something that they don't know about the situation) and so is willing to go in with "all guns blazing" against that one entity they seem to believe is in charge of everything.

I am honestly not certain who is to blame for that: poor schooling, bad genetics, developers who believe software should be "intuitive" (and thereby put the moral equivalent of a cartoon-laden control panel that anyone could believe they could operate on what should have been a very scary buzzsaw in a woodshop)... the cause of this phenomenon is clearly in my "I know that I do not know this" region.

Whatever it is, though, it is simply not fair to state that you, in an analogous situation of being quite far and even fundamentally outside your competence, would act similarly to these confused users; I just don't buy it.

I've found that the tone of the message is often an indication of what the sender thinks it will achieve. The all-guns-blazing "How dare you do this to me?!" is intended to get an immediate apology and admission of guilt, with the probable end goal being a free or discounted service. People who immediately threaten to "go public" with their experience without first trying to resolve it are kind of banking on that sort of response, IME. It's the internet equivalent to "making a scene" at the customer support desk.

It might be just personal experience, but the most foaming-at-the-mouth also seem to be those who maybe realise or suspect they might be personally be at fault, and are writing to reassure themselves as much as blame you.

Those who genuinely don't understand will often to go great lengths to indicate that they've tried, sometimes with speculation on what they might have done or where the problem might be.

Pretty much everyone will occasionally vent their anger at some public contact point of an organisation they're having problems with though, especially if there's external pressure (bank charges, missing SO's birthday delivery, or whatever) on top. The impersonality of email doesn't really help, although anyone who's worked phones or retail knows that some people will rage anywhere.

That's an interesting strategy (attempting to enrage those whose help you are asking): for my part I can say with utter certainly that I bend over backwards to help people who send me the kind requests, and stick to "no, we do not offer that service" for anyone who skirts the "this is making my day unpleasant, I would rather be doing anything but this e-mail" category. :(

However, I am concerned that now in having worded the two responses along the kind/mean access, I have also ruined them as examples of the thing I was actually hoping to demonstrate.

I also get perfectly nice requests from people who still believe that I can help them. I got an incredibly kind e-mail today, for example, from someone asking me to remove software they had previously posted to Cydia (as they were now selling something similar somewhere and didn't want the free version to undercut them).

This request was all-sorts-of-hedged against things that didn't even matter, such as that they had released it under CC and therefore knew I was not legally required to do so, and that they knew I was busy working on other things they appreciated, and that maybe we simply never removed things at all, etc. etc..

However, the real core of the issue is that I do not host their package at all, nor was I even the person they had originally given the package to: they needed to contact the site they were hosting the package with (as Cydia is pretty much just a web browser of third-party content), a repository I happened to know would honor the request immediately.

I was in a similar situation to this recently: I was trying to register my company for physical sales taxes in my county, and figured I needed to contact someone somewhere at the town hall, county center, what-have-you. I went to each of these places, and stated "I am not certain if you are even the person who would be in charge of this, but maybe you will be able to direct me to where I should go" before asking them my question.

I personally believe that that difference is important: I knew that I didn't really know whether what I was doing made any sense, and that I was honest with the people whom I was interacting with that I had no clue about anything related to selling physical products.

I feel like, once you admit that to yourself it becomes much easier to problem solve and work your way into a state of "knowing": in these cases, maybe asking yourself "is this in fact the person I should contact? why am I choosing to contact this person? will I make that person feel awkward if I assume they are in charge of something they are not? who else might I contact in addition to or instead of them?".

I might then go even further and posit that the people who are trying to "reassure themselves" that I or you are to blame are in essence somehow feeling that "not knowing" something makes them powerless somehow, and that they therefore need to be overly confident to "reassure themselves" that they in fact know what is happening.

So, interesting: thanks for responding to my comment! I will think about this more (as I continue to drudge through the last week of hate-mail, which I've been doing ever since I woke up). ;P

I've only been using Chrome for a couple years, but when I try to use Firefox or Safari now, I inevitably start typing searches into the address bar. Every. Single. Time.

I don't think I'll ever be able to use a browser with two text-entry fields again.

You know, searching from the location bar in Firefox has worked for a couple of years now. I use both fields, depending on what I'm currently doing.

You could even change the search engine from Google to something else if you want. It's a field in about:config called 'keyword.URL'. I haven't even needed the dedicated search bar since I switched to DuckDuckGo thanks to the !bang commands. :)

The default has to fit these requirements:

- The OS/desktop manager-specific stuff needs to be there, or else you fragment the overall UX for the computer.

- The browser needs to give immediate access to the actions that a typical user will use all the time.

- Tab mode has won out over separate windows, so you need a visual indicator of what tabs are open and which one is active.

- And the contents of the web page are the contents that you care about right now, or else you're looking for a way to navigate away.

Modern browsers do a pretty good job with all of those requirements. I'm surprised that Bray is even interested in a notional "home" page, since everyone I know, technical or not, either wants their session restored from last time (on startup) or a very fast search page to come up when they tap for a new tab. People don't reset tabs -- they close them or go somewhere in specific.

Best of all, this behavior is all customizable by normal people who never have to type about:config.

No, it doesn't have to be that way, and on mobile devices it often isn't. In a mobile browser a lot of this layering is invisible or only available when called upon by the user. And consider also whether this excessive contextualisation might have something to do with the success of the 'app' model for accessing services, where the device focuses solely on the task at hand and all that framing goes away completely.

Mobile devices have their own UX expectations, and a browser has to fit into that model. You don't get a titlebar plus minimize/maximize/kill buttons because that's not the metaphor that the device uses. If everything on your desktop has a titlebar + buttons, your browser had better have those as well, or else it's a distraction.

Don't break the metaphor. Use local system libraries. Don't make maximizing the window a surprise.

If you want a nonframed experience on a desktop, it can't be the default. You can make it as easy to access as one keyclick to enter kiosk mode. Better provide a clean way of exiting, though, and best is to tell the user how to exit when they enter. (Take a YouTube video fullscreen, FF reminds you to 'Press ESC to exit'.)

I'm not sure we really need all those bars. I mean I use this http://i.imgur.com/0U1yW.png everyday and it's perfectly fine/usable (ok we may need to add an url/toolbar for most people, but at least it merge the tab bar and the title bar).

Edit: It also merges the file/tools/help bar in the left button.

How do you get the Firefox icon instead of the word "Firefox" in the orange drop-down button?

There are probably several addons and themes to that end, here's one that seems reasonably popular:

https://addons.mozilla.org/en-US/firefox/addon/movable-firef...

It looks like it's Pentadactyl (an addon to use Fx with Vim-like key bindings) which provides this behaviour.

Firefox on Windows tends to be more space-efficient than Firefox on Linux or OSX. I don't know why Firefox doesn't draw tabs in the title bar area in Linux (Chrome does), but I heard that it might have to do with some opendesktop standards.

If you want any chance of a Linux app looking at all "native" to the user's currently selected theme, your best bet is to let the window manager draw the window frame, which includes the title bar and minimize/close buttons. Chrome specifically doesn't try to look native in its default state, but also gives the user the option to let the window manager draw the window frame, to fit in at the expense of using more space overall. This is a side effect of X11 and window manager APIs, as well as the proliferation of competing window managers and widget toolkits like Qt, Gtk, etc.

I have Firefox and Chrome on Fedora/KDE, and they actually take up exactly (well, within 5 pixels or so) the same amount of space on top.

Screenshot: http://inst.eecs.berkeley.edu/~tikhon/browsers.png

Interesting. But Firefox still looks more cluttered.

I think the only reason it looks more cluttered is the google search bar, which I never use anyhow. If I get rid of it (I should have done that earlier) and move the buttons around, they start looking really similar[1]. (I also got rid of the home and bookmark buttons which I don't use either.)

[1]: http://inst.eecs.berkeley.edu/~tikhon/browsers1.png

On the other hand, Firefox looks more native--for example, it has the same reload button (but not the forward/backwards buttons, for some reason) as KDE programs.

Also, unlike Chrome, it is pretty trivial to turn off the url bar, saving even more space.

However, my real point is that getting a fairly minimal firefox/chrome (although not as minimal as the example alexis-d gives) on Linux is easy and can, in fact, be done even without any extensions. You could even use KDE's tabbed windows and eschew Firefox-specific tabs entirely, but turning the tabs off in Firefox may require an extension.

Of course, a meta-point is that tinkering around with my window manager is fun :)

In broad terms, this same concern (sedimentation of layers of interface) was one of the original motivations for the MacOS's otherwise unusual single menu bar at the top, rather than separate menu bars for the OS and the application. But that of course hasn't carried over to webapps: while Safari might use the OS menu bar, Gmail inside of Safari layers a new one inside it.

Check out yesterday's browser, and compare it to his FireFox screenshot:

http://screenshots.modemhelp.net/screenshots/Netscape/v7.01/...

Are software UIs really getting more complex, or are they getting simpler?

I'm not sure this is an issue. Every application has a hierarchy of controls at the top of the screen, and the user understands this.

Firefox3 Theme for Firefox4+ and status-4-evar extension solve a great deal of the modern firefox problems and actually make it functional again.

Modern browsers are going to raise a bunch of idiots who aren't even curious about how the page actually works - even "view source" is now buried.

Modern browsers are going to raise a bunch of idiots who aren't even curious about how the page actually works

Historically, 99.9 of the web users didn't cared about "how the page actually works" either.

Btw, are you "even curious" about how your car actually works? Your TV? Your grandfather's artificial valve? X-rays? How milk turns into yoghurt? How epidemiologists study disease? Who was the kind of Sparta at the time of the roman invasion? How the books you read are typeset?

If you answered no to ANY of those, do you consider yourself an idiot?

If no, why should users be "curious" how the HTML page actually works? Because everybody has to have YOUR expertise in the field YOU'VE chosen?

Yes I use my car weekly and learned how to do basic repairs.

I also learned how a TV works decades ago when reading about Philo Farnsworth.

The other items I don't use so I tend to not know how they work.

I didn't start out to learn how to be a web developer, I just got curious how the web pages I used every day worked. Even yogurt has a list of ingredients on the side but "view source" is essentially gone from modern browsers.

Maybe it's because English isn't my native language, but I didn't understand half the worlds in this article.

I mean, 'sedimentation'. What?

I find this a bit odd given that the article is about understandability and accessibility for the layman.

Sedimentation is a geological analogy, as rocks are laid down in layers of sediment and he is looking at the interface layers. It is not normal conversational English but is more of an educated style of writing which is faily common.

Keep at it!

I'm very fond of an article called "Consequences of erudite vernacular utilized irrespective of necessity: Problems with long words used needlessly", won an Ig Nobel prize in 2006 (http://personal.stevens.edu/~ysakamot/730/paper/simple%20wri...).

I agree that my being a non-native English speaker clearly makes the difference in this case, but "more of an educated style of writing" generally means "bad writing" to me. If there's a difficult ("educated") word and an easy word to choose from, there's absolutely no reason to choose the difficult one.

Often you are right, and that article points out some of that. And many people do write badly on purpose, or do not write well by not thinking about it. But simplified English is not going to catch on with English speakers, we enjoy our words too much. The fact that English has more words than any other language is due to constant borrowing from other languages and also constant invention nd reinvention. It is not just educated words, it is full of street words, dialect words, all sorts. And they do convey extra nuances. Look at Shakespeare or A Clockwork Orange for word invention. Of course though there is too much simply bad writing too...

The words in the article aren't particularly difficult for a native English speaker — certainly not 'sediment' — with the exception of 'palimpsest', for which he linked to the Wikipedia entry. With respect, I don't think it odd at all.

This is why I love site-specific browsers. On the mac I make SSBs using Fluid for most of the sites I use throughout the day (email, calendar, google docs, ci server, pivotal tracker, etc.) and then use traditional OS app switching mechanics to switch between them.

As a result, no browser controls: no back, home, or url bar. The window is dedicated to a single site.

And now on Lion, I can make anything fullscreen if I want to get rid of the OS chrome also and focus only on the content of the app/site I'm using.

Does Fluid basically do the same thing as google-chrome -app=<whatever>? (I think it does, but I could be missing something.) This was a big feature touted by Chrome when it came out. You can access it from the UI via wrench menu > tools > create application shortcut (I think it used to be more discoverable, but it is a little hidden now :().

Additionally, you have been able to make browsers full-screen for years, in all OSes (that I've used, anyhow) using F11.

I understand this is an issue, and quite a serious one at that, but how do we solve it for users? As a web dev, we have to assume that the user is decently prepared to use their browser of choice and once on our page "learn" our interface controls, right?

I have trouble finding my "home" button too in Firefox. Maybe I'm just getting older, or maybe things are just getting more complex. Probably a bit of both.

Or they keep moving it :)

I find myself fumbling with the three scroll bars in GMail: Two are owned by GMail, one for the main pane of the UI, and one for the message pane embedded in it. The browser's scroll bar controls scrolling for the contacts displayed on the right side of the window.

A case can be made for this in that all the scroll bars are closest to the content they scroll. But it also points out the unsatisfactory state of having the browser control some scrolling and the active content inside the browser control some of it. This would never fly in the design of any UI library for creating interactive applications.

I'm not sure whats going on in your browser, but I've never seen gmail have more than one scroll bar.

If you have a thread longer than your window, that's one scroll bar. Gmail creates it.

If you have a reply, for example, in that thread longer than a few lines, it is scrolled within a view in the thread.

If you have your chat contacts displayed in the right pane, the browser window's scroll bar scrolls that pane.

That's the case I'm describing.

Wait, people still set a homepage?

In FF, my homepage is Google. In Chrome, I don't need one. Edit: I guess I should mention that I would use the "search" bar next to the address bar to Google for stuff (instead of going "home" to search), but I prefer to leave that one set to Wikipedia.

Yes, about:blank

Reply on news.ycombinator.com