(Archived) Tags vs Categories

[shekala 56]

I am a long time user of Evernote, but am just starting to use version 3. One of the key components of previous versions that I liked was the tree structure of the categories - in once glance, I could see, for example, all deal associated with a given salesman, etc. In addition, I could show the categories in any order I wanted.

Is there anyway to replicate such an ability in version 3? If not, any suggestions?

[crane 40]

You can put the tags (dumb tags, i.e., you have to manually assign them always) in a hierarchy if you like. Just drag a tag to what you want its parent to be.

Unfortunately, there's no way to sort them, except in alphabetical order. I haven't heard EN say anything about fixing that.

Plus, there's not much advantage in using a hierarchy right now, since you can't do an OR between tags. Remember in EN2.2, if you double clicked on a tag, it was selected, along with all of its children? And then you could a search in those categories, i.e., find a note that was in cat A or cat B or cat C? Well, you can't do that in EN3. First of all, double clicking on a tag toggles between select/deselect. So, there's no way to grab a tag and its children. Plus, there's no OR, so even if you manually Ctrl+click on each tag, you get Cat A AND Cat B AND Cat C. Useful, but not useful when you want to see notes that are in some particular category/tag hierarchy. (I do this all the time to see books in my personal reading list - the parent category is books, and I have subcategories for "want to read", "have read", "gave up on reading". I used to be able to double click on books, enter a search term, and find that book. Now, I have to either do the search in each tag individually, or search through my entire notebook, or build some kind of complicated saved search - of which I am permitted a maximum of 32 right now ).

Hope this helps!

[engberg 89]

Searching for tag unions (i.e. notes tagged with either "cooking" or "japan") is supported in the data model, but isn't yet exposed in the Windows client UI. This will be addressed in an upcoming release.

(You can see union searching in the web UI: select one tag, then control-click on a second tag, and change the search operator on the top from "All:" to "Any:")

Thanks for the feedback

[ThomasStraten 0]

Dear Dengberg,

There are problems with the all / any method... Just try a search like

any: tag:one tag:five elephant to try to find the word elephant in both tags.... you will get answers coming from other tags, coming from notes not tagged and getting notes without the word elephant!!!

I sent that today to your beta team with a few other findings

Regards

Tom

[engberg 89]

Correct ... Evernote's search grammar does not allow for arbitrary boolean logical expressions -- the any/all operator changes between a union (OR) and intersection (AND) of all search terms. We are aiming for a more consumer-friendly search behavior that would offer a good balance between expressiveness and ease of use.

While I can come up with artificial examples of full boolean searches that could be conceivably performed, but which would not be expressible in this grammar, I've never actually felt such a lack in successful products like Outlook and iTunes that offer similar (or more restrictive) logical operations.

[madmaxmedia 0]

I have seen other programs that utilize a tag browser pretty well. I think it gives the best of both worlds (if not absolutely perfect). The freeware Mac app SlipBox is a good example-

http://markusguhe.net/slipbox/

[ThomasStraten 0]

Dear Dengberg

(try my example because it does give very unexpected results, might be a small bug in that case).

I am a bit puzzled by the complete change from simple clicking and double-clicking, plus a very simple (although unusual) syntax we (you) had in 2.2.1 to a (sorry to say) very ugly and poor hard typed syntax.

If you wish something really usefull it should be accepting, as minimum, AND OR NOT () and wildcard.

But why such a change? The ease to get notes in is going to be killed by the difficulty of getting notes out in a query.

I doubt that the common user will prefer any syntax when all he needed was some clicking in the category pane, the intersection pane and the search field. Brilliant!

If you wish to have a good example of clear syntax, have a look to DTSearch (inthe help file, there a very clear examples, from simple to highly complex.

The other important thing is that at least the Web version and the Desktop version should look the same (Windows not equal to Mozilla)

Regards

Tom

[crane 40]

Here's another vote for better searches.

According to this forum, it seems that the searching in the web version is more robust, so I went to try it out. Unfortunately, while it does allow me to do the "OR"ing of various tags, that's about all I can do with it.

Consider this example: I have tags for books, "Books To Read", "Books Read", "Books Aborted". I want to look for a particular book, say "Harry Potter". I know it's in one of my tags, but I don't remember which one. I set up the search, but what I get is

find all books with any of tag:"books to read", tag:"books read", tag:"books aborted", "harry potter". This is so unlike the functionality that we used to have, i.e., where we selected promising categories, which were OR'd, and then we got to do an AND with the relevant search terms.

And don't give me that stuff about trying to make it more like google. I know I can do this with google, very easily.

(As an aside, I don't find it reassuring that the EN team has Google search as its goal - yes that kind of search can find stuff fast, but it can't find everything, and there's no way to explicitly search for some things, e.g., "@next". I find the character set supported by this kind of search is really the lowest common denominator.)

[shekala 56]

Is there ANY way to distinguish amongst tags? I don't understand why you are going from a version that had category icons, colors, flags, bold/italic, to one of indistinguishable tags.

Assuming I have a hundred tags, any suggestions on how to distinguish amongst priorities, etc? I am at a loss to understand how to organize tags. Up until now, I could take one glance at my category tree and see what took precedence, what could be delayed, etc.

I hope someone has a suggestion otherwise not sure I can feel comfortable giving up on 2.2

Thanks, David

[crashing_sux 0]

Actually, dengberg didn't compare to google. I believe he was very careful to choose two common applications not known for their great search capability, outlook and itunes. While I don't use iTunes I have been disappointed in outlooks's search many times. Google desktop greatly improves upon outlook searching.

If dengbert compared Evernote's search to google (admittedly, the example most folks would use when talking about great searching) then there would be the expectation to support some of the searching capabitility outlook has.

I am a bit puzzled by his use of iTunes as an example, other than the fact that it is a popular app with weak searching. That is because it doesn't have much to search, all anybody puts in iTunes is media, not the hardest thing in the world to categorize.

It's almost like defending your plans for weak search by saying "look at all of the people who use msPaint, you don't see them complaining about search". Sure this is a little bit of a ridiculous example, but so is iTunes IMHO.

I wouldn't be surprised if the next thing we here is how searching will be powerful and provide full boolean support and intersection capability. I understand the whole beta argument, but what I keep seeing on this forum that is different than other beta forums is that instead of just saying "great idea, we're working on that, might be a few months", what I keep hearing (not always, just too often) from the EverNote guys is "you don't really need that. You can learn to like it without all of those complicated features. Look at how many people use pencil and paper every day, it doesn't have full boolean search and you don't see them giving up on pencil and paper.".

1. If you plan on implementing a feature, just say so, no excuses.

2. If you make an excuse or argument for why we don't need the feature, we will believe it is never coming, EVEN if you then say it is coming. Why would you argue against it if it is coming?

3. If you say it is coming someday, then "it's still beta" is a perfectly reasonable response. We'll get it eventually.

4. If you say it's not needed, then "it's still beta" is a useless response, as we will be left with the impression that the Evernote team doesn't see the value in a feature, and that once it is out of beta the requested feature still will not show up.

Some of the responses are technically correct, but misleading.

PIcture this fictional conversation between me and my GF:

me: I left my lights on all night and now my car won't start. Will you help me jump start it?

GF: You don't need a jump start.

me: Please help me?

GF: I never said I wouldn't help you, only that you I don't feel you need help, you can just walk to work. I was planning on helping you though.

Can you see how that would leave me confused, thinking she wasn't going to help, even though technically she never said she would not help? I feel like we get a bit of that from the Evernote team. Now if the info was given out and specifically said to be a workaround the whole thing would make more sense.

me: I left my lights on all night and now my car won't start. Will you help me jump start it?

GF: I am busy right now and won't be able to help you for 30 minutes. In the meantime, a possible workaround is to just walk to work, or catch a cab.

-Shane

[engberg 89]

Ok, still trying to get a feel for the balance between too much & too little information here. Feel free to ignore my reasoning & examples if you'd prefer to just skip to the conclusion:

The current Evernote search behavior supports global union and global intersection via the Any/All operator, and we don't have any short-term plans to expand this to include arbitrary boolean logic.

As always, this isn't a blanket statement that we will refuse to pursue this later, but revisiting this is a ways down on our priority list.

Thanks

[crashing_sux 0]

Thanks Dave. It it wasn't self-evident, my post wasn't meant as an attack. I see a level of frustration between the beta users and the Evernote team and much of it looks to be plain misunderstanding, but reading the posts I can see how this misunderstanding is occurring in some cases.

It's almost like the apology with an excuse. Everyone knows it's not really an apology, it just makes people angry.

I'm sorry I ate the last piece of cake is an apology.

I'm sorry I ate the last piece of cake but you didn't need it anyway is just making someone angry.

We're working on that feature, it may be a while is acceptable to most folks.

We're working on that feature but you don't need it anyway is just making people angry.

It's not directed at you directly, or any particular person on your team, but I would try hard to give no reasoning for why you aren't going to include a feature if it's coming anyway, just say it's coming eventually. If there is a workaround, give the workaround (I've found a couple of your search examples very helpful) but call them a workaround and say the real fix/feature is coming.

I can understand that internally you have debates on whether or not to add a feature but personally if that debate makes it on the forum without the explicit "it's coming eventually" it will be seen as "we aren't including it.

Chalk it up to human nature I guess.

[shekala 56]

I would like to redirect the conversation back to my original post - namely, is there a way to distinguish amongst tags, and/or are there any plans to make such distinction available -ie color, flags, icons, bold, font size, etc.?

If not, would love to hear from other users how they plan to handle large numbers of tags. Seems to me that the primary point of Evernote is the ability to dump everything into it, and have immediate ability to see relationships, find data, etc. Obviously searching is one approach, but that is just one method to mine data. In fact, in 2.2, I hardly ever executed a search - most of my needs were met by reviewing the category tree and have my eyeballs focus on colored categories, bold categories, flagged categories, etc.

Perhaps I am missing something...?

[crane 40]

PIcture this fictional conversation between me and my GF:

me: I left my lights on all night and now my car won't start. Will you help me jump start it?

GF: You don't need a jump start.

me: Please help me?

GF: I never said I wouldn't help you, only that you I don't feel you need help, you can just walk to work. I was planning on helping you though.

-Shane

Love it!

[crane 40]

The current Evernote search behavior supports global union and global intersection via the Any/All operator, and we don't have any short-term plans to expand this to include arbitrary boolean logic.

As always, this isn't a blanket statement that we will refuse to pursue this later, but revisiting this is a ways down on our priority list.

Thanks

Yeah, but, some of the arguments *against* getting better tags and saved searches is that we *don't need* to categorize as much, because we can simply *search* for whatever we want. But now search is dumber too. What does that make EN 3 then? Just a collection tool? Not a search or organize tool? I'm failing to see how EN3 is distinguishing itself from other clipping tools out there.

[crane 40]

I would like to redirect the conversation back to my original post - namely, is there a way to distinguish amongst tags, and/or are there any plans to make such distinction available -ie color, flags, icons, bold, font size, etc.?

There is currently no way to deal with tags, except to sort them alphabetically. You want them in a different order, put numbers in front of them. Hierarchy is permitted though. Saved searches have no hierarchy.

I have yet to hear anything from EN folks about giving us back the following: ability to manually order tags, ability to use different fonts/italics/colours, ability to use icons, flags.

If not, would love to hear from other users how they plan to handle large numbers of tags. Seems to me that the primary point of Evernote is the ability to dump everything into it, and have immediate ability to see relationships, find data, etc. Obviously searching is one approach, but that is just one method to mine data. In fact, in 2.2, I hardly ever executed a search - most of my needs were met by reviewing the category tree and have my eyeballs focus on colored categories, bold categories, flagged categories, etc.

Perhaps I am missing something...?

Other threads have discussed the concept that EN 3 is more of a clipping/grabbing tool. Not so good for organization. The CEO himself is not much of a tag user; he is impressed that he can now find the use for 2-15 tags.

No, you're not missing anything. It's just that 2.2 is so much better at organizing, data mining, etc. Stick with 2.2. Maybe one day EN 3 will catch up. If it does, it won't be for months, maybe years.

[shekala 56]

Very unfortunate, indeed. I have been looking forward to web syncing for so long, and when they finally do it, they cripple the program. Amazing. Are they really so blind to how important the category functionality in 2.2 is to the existing user base? Major disappointment. Talk about 1 step forward, 2 steps back...

[crane 40]

I am a bit puzzled by his use of iTunes as an example, other than the fact that it is a popular app with weak searching. That is because it doesn't have much to search, all anybody puts in iTunes is media, not the hardest thing in the world to categorize.

By the way, you can actually get quite sophisticated searches in iTunes, but it requires using several smart playlists. For instance, create one playlist that collects music based on Rock OR Pop OR classical. Then, create another playlist that collects music based on Playlist1 AND artist=pink. Or whatever.

To get similar functionality in EN, we would need the ability to used saved searches themselves as search terms. But it's most definitely not elegant. I'd rather be able to do a search like "(rock|pop|classical) pink" as a search

[crashing_sux 0]

Ok, now that's good for a laugh.

[ruudhein 29]

hear from other users how they plan to handle large numbers of tags

Over the years my tagging has changed from being very strict and limiting to attaching any keyword which at the time of capture pops up in my mind as The Keyword(s).

Thus on del.icio.us, in TheBrain and in within short time since really starting to use EverNote3 I have huge amounts of tags.

I'd love to see EverNote become as smart as the del.icio.us Firefox extension where it's easy to see with which tags a tag appears.

For the moment I don't expect so though -- and it doesn't take away from my tagging. I plan to use them to find things; tag:USA -soccer

have immediate ability to see relationships

For that I use TheBrain. EverNote is, as far as I know, not meant to help crystalize relationships in data. It *could* ... but it wasn't the purpose, right?

My take on EN is that it's meant to store the little snippets and notes we come across. "Brian tel. 123-4567" for example... But no way to see that Brian has 2 kids, that he used to work for IBM and is married to Suzan, the girlfriend of your wife.

So for the first stuff I use EN. List making, initial capture, data storage. For the latter I use TheBrain where I can make things as complex as I want and express relationships visually, with link types, thought types, etc.

Leading into:

I hardly ever executed a search

I search *all* the time. Only reason I tag is that it helps with some organizing, some searches, some saved searches. But I *search* almost all the time.

I help that along by "tagging" inline; adding keywords I know I might use in my search to find this particular note.

Usually I'll CTRL + END to the end of the note, CTRL + L to go left aligned (in case I'm at the end of EN's right aligned source reference), CTRL + I to go italics and then I add words. Any words.

So for the example of Brian's telephone number I might add coworker friend mobile cell phone

The 4 (or 5) people I've got hooked on EverNote since 2005 are searches too. In fact, one of them was extremely reluctant to go the whole complicated route until I explained that there was no *need* to use the categories, as they were called back then.

I tell you, the amazing instant search is THE selling point of EN to me and was THE reason I bought the product (back then to support the company even -- remember when they were doing so-so?)

I'm failing to see how EN3 is distinguishing itself from other clipping tools out there.

The powerful instant search.

I think most people would rather add a tag or a word to a note about a book then type a complicated query in order to find a specific intersection listing all books (1) they want to read (2) but don't have a blue cover (3).

Also, boolean search can be very powerful, going as deep as stating the word "book" has to be within 6 characters or words on the right or left side of the query etc. Many legal databases still work this way. You have to be very comfortable with the concept and its expression though.

Meanwhile most people type short-ish queries. The 2 and now 3 word queries on Google. And when those queries aren't answered only 10%-ish switch to advanced mode; most people will revise or refine their query. ... Exactly as I use EN and exactly how EN is build to interface with us

[crashing_sux 0]

Also, boolean search can be very powerful, going as deep as stating the word "book" has to be within 6 characters or words on the right or left side of the query etc. Many legal databases still work this way. You have to be very comfortable with the concept and its expression though.

Can you expand on his, give some examples? I thought George Boole's work just left us with AND, OR, and NOT.

[shekala 56]

My take on EN is that it's meant to store the little snippets and notes we come across. "Brian tel. 123-4567" for example... But no way to see that Brian has 2 kids, that he used to work for IBM and is married to Suzan, the girlfriend of your wife.

Perhaps explaining how I use Evernote will help dig into this issue further. I have two laptops - one is used as a standard desktop (with external monitor, keyboard, etc) and I use it for keyboard intensive work like spreadsheets, email, etc. My other laptop is actually a Tablet PC and I have it open in the tablet mode 99% of the time, sitting on my desk as if it was a yellow pad, open to Evernote, of course.

As calls come in throughout the day on various deals I am working on, I take handwritten notes in Evernote. Then, a couple of times per day, I categorize each note as a certain deal.

Bottom line - I end up with a category tree with salesmen at the highest level, and their respective deals as subcategories. By using colored categories, bold/italic, flags, and manually ordering the categories, I can quickly look at my category tree and see which deals need immediate attention (generally those in red), etc. In other words, my list is not simply the names of a hundred deals, but provides visual clues as well.

Lets face it - as good as Evernote is, searching on handwritten notes is not that accurate (especially with my handwriting), and since all of my notes are handwritten, I rely on the category tree for searching, then simply manually scroll through each note in the required category to review the latest information on a given deal.

Seems to me this is going to be much more difficult in 3.0 given the lack of visual clues associated with Tags...

[ruudhein 29]

Can you expand on his, give some examples? I thought George Boole's work just left us with AND, OR, and NOT.

Basically, yes. But most models have been expanded with proximity search. So a legal search engine can do something like evernote /3 versus /s onenote which translates to a search for evernote with the word versus within 3 words and the word onenote in the same sentence.

Proximity search is an early version of phrase-based matching but often more precise in its expression.

Seems to me this is going to be much more difficult in 3.0 given the lack of visual clues associated with Tags...

I see where you're coming from. I too have considered putting forth a feature request for either custom sorting or, perhaps simpler, an additional pane in the sidebar where I can custom sort some categories.

But I rather play with what's there and make it work within those parameters. Perhaps using @ to force a certain sort, for example.

Why?

I don't like having to wait for a feature -- and how many people have requested things for 2.1, 2.2 only to see features they had grown used to removed and their requested feature still not implemented?

By "making due" I feel I make myself less dependent on the company...

[ThomasStraten 0]

Dengberg said:

The current Evernote search behavior supports global union and global intersection via the Any/All operator, and we don't have any short-term plans to expand this to include arbitrary boolean logic.

Dear Dengberg,

I suppose your system designer has a PhD. How come that the most important feature of your search engine is THAT limited (only all/any) when everybody who has experience in programming knows that it is absolutely not sufficient for, inclusive, very simple searches?

The more I Betatest and the more I feel afraid....

Tags are not working as they should, and are bugged.... (thus, EN3 not very useful and misbehaving)

Search is STRICTLY limited to all/any (absolutely unefficient except for very simple databases)

This leads to only one thing: VERY SIMPLE DATABASES... (basic tag is just a keyword. Just as in the less efficient databases or pims)

Is this the future of EN3? A throw-everything-into-it for basic users (business man with contactlist, places to dinner, last-seen-stuf on the web that seem fun)? And that would be it? IS this Phil's view?

Because in that case, it does not make sense to make an effort improving Beta3 up to the level attained by 2.2.1, that would be a waste of money... And it helps me understand a bit more the rush to put Beta 3 ASAP even with flaws, the words of Phil, the videos, etc...

Very simple databases (notebooks)? Answer that please Dengberg. We ought to know. We deserve to know (I humbly think so)

Regards

Tom

[engberg 89]

Sorry, only a Masters ... and from a state-run school, no less. ;-)

One of my favorite applications of all times is Emacs. It's the second thing I install on a new computer (after Firefox), and I use it every day. Emacs lets me do absolutely everything via the keyboard, and all aspects of it can be customized by writing scripts in Lisp (arguably the most powerful programming language ever). Control-x-2 splits my edit buffer in half, Control-x-o switches between buffers, Esc-x-h-e-x-l-m-o-d-e switches the editor into binary editor mode, etc. I absolutely love Emacs, and I've been using it for around 16 years.

I also think Emacs a terrible application for 99.999% of computer users, and I've never recommended it to another person in my life. It's unintuitive (I'd say anti-intuitive), the learning curve is horrible, and a lot of common things don't work right when you install it (e.g. printing).

We're trying to find the right balance between the power of something like Emacs and the ease of use of well-designed consumer products like many of those from Apple, Sony, etc. It sounds like you don't think we've found the right balance yet, and we appreciate the feedback.

Thanks