I have written an editorial/analysis about the spatial metaphor and surrounding debate. My perspective is one I’ve not seen represented much in the debate so far.
I have written an editorial/analysis about the spatial metaphor and surrounding debate. My perspective is one I’ve not seen represented much in the debate so far.
Thanks for the editorial…
I am amazed by the study about short term use of desktop and long term use of hierarchy. Being a long time computer user, this is still the way I work for probably 50% of my files.
I agree on hybrid model… I am using nautilus navigation mode but I would love a “Save on Desktop” function which would instantenously show me the desktop so I can spatially place the file I am saving where I want it to be on my desktop – Instead of having to save in Desktop folder and then having to move it so not all files/icons appear on left top corner of my desktop !
i like the plinko effect. (price is right)
just drop files and see where they end up..
then google search them later.
“A”
http://www.cbs.com/daytime/price/games/plinko.shtml
I’ve tried using finder-like spacial interfaces where folders each have their own identifiable and predictable screen location and size. This worked OK for a little while, but I can only tell the difference between so many before they all start looking alike and paths become confusing. I might know that what I want should be in the top left, but exactly where in the top left? And what is the path I should take for a folder in a folder in a folder, etc. My brain isn’t big enough to handle all of this.
On the other hand, I’ve found that trees make this much easier. You could usually draw a picture showing where things are located in a tree. (In fact, I found myself usually relying on this mental visualization as a crutch in the finder-like interface.) I suppose you could say this is its own kind of spacial. On top of that you have relationships and hierarchy to give you clues that can be more quickly and easily seen all at once or tried one at a time without the same tedium.
the column browser from next step / mac os x works pretty well.
it lets you see the hierarchy of the system, and lets you drag and drop between levels.
I don’t care for the spatial paradigm because I do not believe it scales well to a large number of locations. For the extreme example, consider what the web browser would look like if it were implemented spatially. Every time you clicked on a link it would pop open it’s own window with it’s own unique size and screen position. Part of the problem is that at some point this location is completely arbitrary. The location of a particular window is only going to be meaningful for a few sites that you hit often. Another problem is that you end up playing window whack-a-mole and hunt the wumpus as your eyes wander around the screen. Two thumbs down.
Michael
Back in the day 😉 I was a mac-only guy. Bought a Mac 512K when I got out of school. _Loved_ the spatial finder.
But I took on a large programming job using A/UX, which was Apple’s first Unix OS with the MacOS user shell over the top. I quickly found out that the finder was worse than useless for navigating through the System V filesystem. But I was still completely happy with it in my home directory, where I had organized things in a flat structure to best use the strengths of the finder.
My point is that you have to use it only in places that make sense, and that you can’t easily use it in places that don’t fit the spatial metaphor well.
My
A very carefully laid-out argument.
One addition I would make: Christian Paratschek’s quotation about how users can’t find the email attachment they just saved is spurious; that issue has nothing to do with the interface of the file manager — sure it’s annoying, but it’s annoying behaviour on the part of that application program, and would be the case no matter what the file manager’s UI.
Its not only a good article, but it actually has something besides opinions and ranting! What a concept!
Seriously though, those studies are very interesting. Especially the one showing spacial use for short-term storage and heirarchial for long term, sort of like the difference between ones desk and ones filling cabinet. UI design gold.
i do not think in a ‘spatial’ way. i could not force myself to use this system.to me it appears to be a way to promote clutter. then again my desktop is completly BLANK and i hav yet to find one person who can sit down at my computer and figure out how to do anything…. i just consider it a form of security.
I don’t understand the problems people have with the spatial metaphore in Nautilus. It’s not like you can’t use the browser if you want to. Also, if you don’t like window clutter, rightclick on a folder and select “browse folder”. Having both methods available is increasing choice which is what OSS is all about. If you absolutely don’t ever want to use the spatial metaphore, just switch it. It is available as a checkbox in nautilus prefs. I for one have my most used folders on the desktop and most of them are not more than one or two folders deep so the spatial way is very convenient not to mention faster. If i need to browse, I just click the browse filesystem icon that i added to the panel. This is a rediculous debate now with gnome 2.10 out the door. When they first switched to the spatial way it was an issue because there was no easy way to move back but it is null and void now.
Use what you like, If you like the browser, use it. If not, leave it spatial. Both are available in Gnome. However, for new computer users, I think spatial is the way to go. Maybe not for Windows converts but for new computer users yes.
The problem that I have with spatial is not just about the fact that (as hard as I have tried to like it) I hate it. Of course I know how to avoid it! My major problem with spatial is that it is being presented to end users as the default and is shaping their perceptions of what Linux is like, and it stinks. And the whole idea of spatial is based on a bunch of unsupported, theoretical, pie in the sky, ivory tower mumbo jumbo.
people work well with spatial concepts when it comes to relatively small numbers of items (the sort that we can hold all at once in our mind, e.g. 7+/-2). this is why the early Mac spatial experiments worked so well: data sets were limited (largely by disk space and RAM/CPU capabilities).
hierarchies scale better (in part because they are “reverse-engineerable” by applying logic to the structure and therefore have a longer shelf life than our memory), but even they top out. i personally believe those those limits were achieved on PCs a few years ago.
i resonate with the “hybrid” approach that the author of this essay talks about in that there is no One Right Way for all information retrieval interactions. in fact, we seem to be in need of other ways of operating that scale beyond what hierarchies are capable of.
the other Aaron in this thread mentioned a Plinko concept. this sort of “drop it and search for it later” can be very powerful and can be made to scale up very effectively. (you can also completely screw the pooch on it; it requires a bit more than simply “search a metadata index”)
there are several very interesting methods of providing input (in one views “Save As” as an input method) as well as retrieval when one allows for non-spatial, non-hierarchical methods.
in the Appeal Desktop Project, we are working on a number of these concepts. due to not having a lot of pre-existing examples to copy from, it’s currently a matter of “design something new, test it out, return to the drawing board and repeat”. it will be interesting to see which ideas ultimately survive user testing at the end of the day =) when we have that information we’ll be releasing demos for everyone to try out, and at least some of these will be making their way into KDE 4. personally, i’m rather excited by it all =)
so …. IMHO, the future of desktop data retrieval is neither spatial nor hierarchical. those are methods best used in limited contexts and while they will continue to feature in certain applications, true information retrieval systems will be where more and more of the interface heads. =)
I organise my files like the article describes. I arrange temporary icons into groups on my desktop and use hierarchies for storage, projects, etc. However, the hierarchies of folders are deeper and more complex than I’d like them to be because there’s no other way to arrange files except in folders. One way of getting around this is using a file manager that remembers the spatial arrangement of icons, so you can arrange them into groups within a folder, adding some more structure without creating a deeper and more complex hierarchy. That’s the spatial metaphor. But that doesn’t allow for sorting operations. I think it would be better to allow use-defined categories within a folder. The contents of a folder could be split into labelled sections that can be sorted however the user likes, and internally sorted however the user likes. It would also be nice to put other information in the folder, like due dates, to-do info, notes, etc, that are displayed inside the folder itself, rather than having to be put in a text file within the folder.
As for the “spatial debate”… This whole area of HCI is just nitpicking. Give the users what they want. The fact is, HCI moved on; the quantitative pyschologists, concerned with shaving milliseconds off tasks performed by someone sitting in front of a screen, were replaced by more relevant research like workplace studies a long time ago. When you actually look at what people do with their computers, instead of timing how long it takes them to move the mouse to different parts of the screen, it becomes clear that the BIGGEST time saver is letting them customise their environment and set up their own workflow for their needs. We make it needlessly difficult for users to do that by putting these “expert” settings out of the way in bloated preference screens, removing their context.
I’m frequently baffled as to the bad rap heirarchies get in these discussions, given how central they are to how most people think. Most Americans will say that cats and dogs are different from chickens and cows, jazz is different from country and western, and of course, chick flicks vs. guy flicks.
I have a search system that searches my entire home directory every day allowing me to find any text, pdf or oo.org file by keywords in the content in under a second. However, it is not particularly useful without my file path telling me if that file is something I wrote or something I downloaded. And if I did write it, the file path tells me which project I wrote it for.
Which is why I don’t buy the claims that naive “plinko” will be the wave of the future. (I really need to finish my article on this.) Google makes it fairly easy to find ANY webpage relating to a topic, but somewhat more difficult to find A SPECIFIC webpage and frequently flounders dealing with multiple meanings of a word.
Now what I think would be really cool, would be virtual folders combined with some automatic statistical classification system that can infer this file is work-related, this file is notes for the novel I’m not writing, this file is an invoice, and that file is downloaded reference material. But that would also require some degree of user-work.
To a large degree, I think the spacial vs. browser debate is primarily trivial window-dressing.
I have written an editorial/analysis about the spatial metaphor and surrounding debate. My perspective is one I’ve not seen represented much in the debate so far.
>
>
My perspective on the the spatial metaphor and surrounding debate can be summed up in two words:
Who Cares?!?
If you like the spatial metaphor, use it.
If you don’t,then don’t.
What’s the big debate here about anyway?
Another not terribly-well recieved instance of the UI “Community” trying to shove their “suggestions and opinions” down the throats of the the Free Software/Open Source movement maybe?
FTA: “Revolutionary innovation is difficult and often ill-advised in this space, given strong de facto standards. But there’s plenty of room for incremental innovation and innovative new features.”
Right on. User interfaces are still far from perfect, so let’s continue experimenting with revolutionary different ways. But don’t forget that people have gotten used to some common UI designs. So if you come up with something new, only make small incremental changes (and revert when people don’t like it). Or keep the ‘known working’ as default, with an option to try the new style.
I think the author (and commenters) are right, that the spatial approach just doesn’t scale well: with few files/folders, you’ll remember where everything is. With many files, everything will start to look alike.
The more data people collect, the more search/database style approaches will be needed. I think simply like “random fuzzy search a la Google” combined with “simple, effective ways to confine the search space”. Enter “audio” somewhere -> search confined to mp3/ogg/wav/etc. Recent file? -> Search confined to files updated recently. And so forth.
Just a tip: I enormously like the ‘QuickBrowser’ feature as included in recent Knoppix releases. Click somewhere to start, then move trough directories by just moving your mouse (a la Windows startmenu). Very quick navigation, and saves many mouseclicks. Just boot a Knoppix CD to see what I mean. Adding a context menu (right click) to that, would make it very useful for day-to-day file management. Anyone know exactly what software makes this Knoppix feature happen?
But no matter how good tools you have, accept it: some people never learn how to organise their files, and never will. When a friend tells me “I know I’ve got that file somewhere, but I can’t find it anymore”, I always laugh. If you can’t find it back later, what’s the point in saving it in the first place? When your harddrive fills up, you don’t need a bigger one. If your harddrive is full, you’ve got too much crap stored on it – time to clean up.
“Just a tip: I enormously like the ‘QuickBrowser’ feature as included in recent Knoppix releases. Click somewhere to start, then move trough directories by just moving your mouse (a la Windows startmenu). Very quick navigation, and saves many mouseclicks. Just boot a Knoppix CD to see what I mean. Adding a context menu (right click) to that, would make it very useful for day-to-day file management. Anyone know exactly what software makes this Knoppix feature happen?”
i have no idea as to what lets it do that but i know on windows if you add the “desktop” tool bar to the task bar you can do the same thing, it even lets you use the context menu. can’t say i’ve ever found it useful but the idea is out there
“I have a search system that searches my entire home directory every day allowing me to find any text, pdf or oo.org file by keywords in the content in under a second. However, it is not particularly useful without my file path telling me if that file is something I wrote or something I downloaded. And if I did write it, the file path tells me which project I wrote it for. ”
thats where metadata comes in, we’re still faily early in the whole “search as a file management paradigm” but once you have things like beagle and spotlight and whatever microsoft is calling there’s it’s not too much of a jump to having those programs display metadata like what user/computer wrote it. and when saving files instead of say choosing the directory and then typing in the name you would choose to associate it with a program and type in a name so when it comes up in a search underneath it says something like
“MyNotes.txt
attached to: projectfoo
written by Axel on Computer”
i mean like i said it’s early there are kinks to iron out and search kind of lack the ability to “discover” things you haven’t used in a while. i mean you can always give it a query like “accesed longer than 1 year ago” but that’ll bring up a bunch of non relevant system files, kinks.
I’m fine with file managers defaulting to either method, as long as there’s a way to change it. Sometimes I use multiple windows, but most of the time I prefer a hierarchical file manager.
Back in the early Mac days, I remember laying windows out in different positions when I was new to computers, so that each window wouldn’t overlap any other as I traversed the file system. That quickly got old, and the only reason it worked at all was because of how few icons were in each window.
The only thing OS X Finder seems to do differently than the other file managers is bringing the same window to the front if you open it again. That’s really not a problem in Windows. So you have more windows to close when you’re cleaning up, no biggie, you still see what’s in the folder just fine no matter which one you switch to.
And the Gnome file manager has worked the same as it always has. They just, by default, removed the –browser argument from the shortcut to Nautilus. First thing I did when I upgraded to that version was notice it wasn’t the way I wanted it and typed those 9 characters back in. Took all of 5 seconds. Meanwhile everyone’s switching WM’s and complaining everywhere. Its really not that big of a deal.
Also, I find having a clean-looking desktop much more important than instant access to hand-placed icons. Even when I used to use the desktop for icons, I preferred them to be in a straight line, automatically laid out by type.
But now I find it more soothing and clean-feeling to remove all built in icons from the desktop, except occasionally a single folder, or sometimes downloads scattered about temporarily when I’m feeling especially lazy, only to be cleaned up when I’m not so lazy.
On Windows, OS X, and Gnome, I use a quick launch, panel, or dock on the left for all my favorite programs, where it can be accessed easily at all times, and isn’t taking vertical real estate. But I sure miss the taskbar when I’m using OS X.
Again, defaulting to spatial is fine if it helps new users, but they better get used to hierarchies anyway, whether its tagging metadata for searches in Spotlight or storing personal files in Explorer, it is, and probably always will be, a common pattern used in computing, and for good reason.
thats where metadata comes in, we’re still faily early in the whole “search as a file management paradigm” but once you have things like beagle and spotlight and whatever microsoft is calling there’s it’s not too much of a jump to having those programs display metadata like what user/computer wrote it. and when saving files instead of say choosing the directory and then typing in the name you would choose to associate it with a program and type in a name so when it comes up in a search underneath it says something like
I guess one of the things that is underestimated by all of the “search as file management” is that file heirarchies are a form of personal folksonomy. For example, most of my projects involve separate folders for report drafts, research data, presentations and correspondence.
In other words, the process of attaching files to project categories is still likely to look a lot like saving a file to a folder, and require about the same quantity of “work.”
Isn’t the point of WinFS, Beagle, Spotlight, Google, Storage, etc the ultimate goal of filesystem ignorance?
Aren’t we hoping to one day not have to click through layers of folders and rather find the file we want instantly?
I was very interested by the finding that people use their desktop for recent activity and hierarchies for archival. I find that i also do this.
What if a filesystem were to auto archive things for you? Maybe there’s a “Auto-Archive” (or some snappier name
) on the desktop or on the OSX Dock where when a user is finished with the recently used or recently downloaded files, drags (or otherwise marks) to this “Folder” which calls an auto-organizer that organizes everything in a heirarchy based on extension/metadata.
This hierarchy is merely a formality though, because the user will more likely access these files through Spotlight/Beagle/WindowsSearch. Then maybe the search results have an automatic behavior where double clicking on the result (which should be fairly large to prevent misclicks) moves the file back to the desktop.
Perhaps this last bit is counter-intuitive and should rather open/execute the file as usual. The main idea is the auto-organization.
Starnix is totally right and I’m still amazed that so many people don’t see it. Spatial is not about a specific filemanager functionality, it’s merely a side-effect of object oriented thinking.
It makes sense, that doubleclicking a folder will open an object that represents just that, a folder. Tools like a browser are their own objects, so they can happily coexist. This has nothing at all to do with defaults.
It amazes me that people still claim that “spatial fans” are the ones who are asking for an all-or-nothing solution, when it’s exactly the opposite! “We” believe that a browser is better when it’s a real browser and a folder is better when it’s a real folder. Once you “merge” the two, you don’t have real folders anymore. There is no such thing as a hybrid (although the OS X finder might be a good model for an alternative browser).
Next, the claim that hierarchies scale better (Aaron Seigo) is interesting. Because there is absolutely nothing about the spatial concept which would render hierarchies useless. If you believe that a tree is easier to navigate, you can perfectly implement it in spatial folder views (the classic Mac finder did just that). If you want to, you could create a dedicated treeview tool, no problem! Even if you don’t want to actually use the browser.
In the near future, we will have sophisticated desktop search utilities. But does this render folder objects obsolete? No, because folder objects have nothing to do with locating data, they are all about actually using the folder, once you found it (by whatever means). I often use the location dialog to quickly bring up a certain folder. That is somewhat geeky and not a great argument for spatial folders, but it shows how well they work in conjunction with other means of accessing target folders.
Maybe you never actually want to use a folder, but it’s invaluable to have in case you do. If all you ever want to do is to locate and access documents, then there will certainly be better methods for this in the future.
When it comes to working with folders (creating new documents, moving/renaming files, opening documents, etc) nothing beats the simplicity of a straight spatial representation of a folder. The scale argument doesn’t apply here, because not many people use to work on 100 things at the same time. Usually I work on at most three projects in parallel, each on their own workspace and with the appropriate folder windows still hanging around. I never have to locate anything, because it’s right in front of my face!
I believe in a future where the desktop does not consist of all-in-one applications anymore, but of lots of dedicated tools and objects working together. Yes, this will mean that we occassionally have to open more than one window, but that’s the whole point of it.
It’s the only way we could ever hope to archive the same kind of simplicity and interoperability we are used to from the Unix CLI tools (minus the cryptic commands). We will need more windows, but the windows will become simpler in turn. Your desktop will become your IDE, not a lifeless shell for applications.
You don’t have to agree with this vision, but please try to at least understand the ideas behind it. Most comments about “spatial Nautilus” simply don’t do it justice.
But what if the Desktop is the universal workspace. Then files you don’t use as frequently as accessed via a search tool.
Sure you don’t want to have to run a search query every time you need something, but searching might be a faster way of finding archived data.
Interestingly, I had the same thought of an auto-categorizing “Archive” folder at some point. But then again, you said:
“This hierarchy is merely a formality though, because the user will more likely access these files through Spotlight/Beagle/WindowsSearch. Then maybe the search results have an automatic behavior where double clicking on the result (which should be fairly large to prevent misclicks) moves the file back to the desktop.
Perhaps this last bit is counter-intuitive and should rather open/execute the file as usual. The main idea is the auto-organization.”
If the hierarchy is merely a formality (and I’d agree), then you could simply create an “Archive” folder yourself and throw everything you don’t currently work with inside there, using query tools to locate it if you need to. So you don’t even need any extra-technology, it will all come together.
This still sounds to me like the most likely outcome of this “metadata-revolution”. I doubt that I will ever stop using direct file/folder representations completely, because no matter how efficient the search tools become, having the file directly in front of me on my desktop is even more efficient. Why search what’s already there. 
I see the biggest advantage in the possibility to move all the cruft out of my way which I don’t currently need (and still find it quickly), so my actual workspace could become a lot more efficient and much less crowded.
Aren’t we hoping to one day not have to click through layers of folders and rather find the file we want instantly?
A more probable outcome is a system that forces the users through multiple attempts at search-engine fu to either hit the file, or pick the file from a reasonable list of options. The reason why I don’t find a full-text search to be that useful for finding files on my hard drive, is because it is hard to get a set of search parameters that hits the sweet-spot of 5-10 relevant results. Instead, what usually happens is I get results that are not relevant, or too many results.
What if a filesystem were to auto archive things for you? Maybe there’s a “Auto-Archive” (or some snappier name
) on the desktop or on the OSX Dock where when a user is finished with the recently used or recently downloaded files, drags (or otherwise marks) to this “Folder” which calls an auto-organizer that organizes everything in a heirarchy based on extension/metadata.
Programs that move things automatically for the user are generally bad. (For example, Microsoft’s missing menu items.)
I agree.
That was in fact the main point I was trying to make: Spatial Nautilus cooperates very well with other technologies.
Personally I think that there are few UI issues more frustrating than multiple windows open on my desktop at the same time as they are rarely organized and are often hidden. They also create this chlostrophobic (sic) feeling that makes me want to pick up my monitor and throw it out the window. I do think that it is important to be able to ear-mark a window as to be able to switch back and forth between tasks (tabs along with book-marks do an great job of this.) I personally feel that for myself any way that a browser or commander style interface is the best non-commandline interface we have. The spatial paradigm makes the a computer much more complicated and cumbersome than it needs to be. I think that one application per desktop is more than enough for me as personally I am usually only using one app at a time and I do not want to be distracted by others. This is why, for instance, Amarok has such a great UI, I can have one desktop dedicated to my music, while konq is great so that I can have another desktop with one tabbed interface dealing with my files and the Internet (what difference is there between these two anyway?) I think that Google and the Web demonstrates how a search technology is very useful for getting to the right higherarchy while the browser metaphor is the best for taking your search form there. I don’t want the last ten web sites that I have visited sitting underneath my browser and I don’t want the last ten directories sitting under (or worse next to) my file manager. One app per desktop and then let that app manage windows within itself via tabs or whatever else is appropriate.
A window manager is one of the least effective tools to balance tasks for me.
Sorry about the rant, and remember what say applies to me only, although I do think that sane e.g. KDE like defaults are important.
Maybe spatial works better with “concepts” rather than folders.
Or we could simply ditch them and go for ZUIs.
Why can’t we have the option to single-click to browse and double-click to open a new window? This could work in both file managers and in web browsers. This way the user can easily create a new window if they want or not if they don’t. Give people all of the options and let them decide for themselves.
The article is bland. The author suggests the Nautilus camp adopt the OS X’ way of managing files for absolutely no convincing reason. Where are the studies that prove that OS X’s way presents the most optimal solution for managing files?
What’s further frustrating is that while the author is quick to deal blows on his supposed weakness of the spatial paradigm, Nautilus in particular, he did not provide a possible solution to the “Nautilus Spatial Crisis.” He proposes that we adopt OS X’ Finder system because, well, it is Apple and they must have conducted empirical studies, and the Finder system is neither navigational nor spatial.
That is the author’s conclusion. A hybrid system is the answer because Apple uses one. The author has two qualms with the spatial paradigm. The first is that it opens too many windows, and the second is that there aren’t enough studies to prove it is a better way of managing files.
His criticisms have holes. A user need not have many windows open as the user navigate through the system. There are functions in Nautilus that enables a user automatically close the parent window as a child window opens. So his first qualm is rendered null and void.
As to his second issue, I haven’t seen any studies that bolsters either the spatial, navigational or hybrid paradigms of managing files. So, why does he single out the spatial paradigm? If the author at least attempted to provide a solution to the problem he sees, I wouldn’t be a disappointed as I am now.
This guy is on the right track. I hate the spatial browsing and so does every single person that I have ever seen try to use it.
I honestly tried it out for about 3 months thinking “yeah, ill get used to it and then it will all be better”…didn’t happen.
I find it amusing to recall how Windows 95 defaulted to this option of “open files and folders in a new window” or something like that. One of the first things I did on install was to kill that option.
And when I want to browse my filesystem, do I open My Computer?
No, because it’s hopeless! I open a proper Explorer window with the handy tree view down the left.
The crucial thing about this is that I can go anywhere on the system when I want; I don’t have to back up from the current directory and drill back down.
Which is why Konqueror is my favourite; it may have many issues, but it has a beautiful integration of tab completion into the address bar, and a proper hierarchichal tree view panel (even if it doesn’t always work perfectly).
Don’t get me started on the GNOME, Firefox or Thunderbird context dialogs that make it impossible to navigate anywhere in a reasonable period of time.
Okay, that’s not so much because they’re spatial, but as far as I can see things go to crap once you start trying to hide the filesystem hierarchy.
It’s obvious that you are not the target reader for this well written article. There are people who are interested in this. You should follow your own advice and stop complaining. If you don’t like these kinds of articles, don’t read them 🙂
It is bad enough that they thought it was a good idea, it is a much worse idea that you have to sort 3 menus deep and uncheck something that mankes no mention of spatial anything in order to dissable the stupid feature.
The spatial browser makes me want to throw something. Even after looking around quite a bit for how to turn off the feature I had to ask in an IRC chan before someone told me how to kill it.
From all the people I have talked to about it, the general consensus seems to be to get rid of it.
Just another example of Linux developers thinking they know what is best for their skewed perceptions of “the end user”
If linux were on 85% of the desktops a change (mistake) like this would leave people outraged.
I can bet money that this will eventually be changed back, maybe not in the next few years but eventually. Why not just skip ahead and fix it now?
Thank you OS News. As a HCI researcher, I have to say that this was one of the better usability articles here. The author shows reason by asking that evidence be provided to support assertions and also trying to make people realise that without this evidence, their opinions are not always correct (in fact, quite often they are plain wrong).
Excellent read, and quite heartening!