We only put any trust in these, if the identity of every author is human and based on “Today trust is based on digital signatures and classical questions of authentication, integrity and non-repudiation, all backed by a chain of trust traceable back to some well-known certification authority. In some contexts, this hierarchical, binary view of trust is adequate.”

Because rating, voting, scoring is only a trust indicator if the identity of the author is human and has a good reputation. Reverse Citation is no indicator of trust, because you never know if the citation does express trust or mistrust in the subject matter cited.

Reputation is the result of identity and humanity. Or would you trust a (web)robot, just because it is consistent in it’s spamming comments on the web?

Which all brings us back to the Hierarchy of certification of identity and trust that might be assigned to it.

The only thing we can change is the well know root of the certificate chain. There is no need to only trust commercial root cert or public global root certs. I can trust the (self signed and free) root cert my running club does issue and uses for signing its members, etc.

But I’ll give you an anecdote. Years ago, maybe 1989 or so Lotus 1-2-3 had a limitation in how many unique styles could be used in a spreadsheet. Although there were not user-visible named-styles, each unique combination of typeface, text color and background color was stored as an index into a table. Think of it like a pallete. The table was limited to 255 entries. So a user could only have 255 styles in a spreadsheet. Now this sounds like high number. How many styles do you typically use in a spreadsheet? 4? 10? But you would not believe how many calls support received from customers who exceeded this limit. What were these customers thinking? What were they trying to accomplish? Was directly manipulating cells to create 255 different attribute combinations a useful thing for them to do? I tend to think it was not, that they their goal was something else, but the tool failed them and forced them to deal with minutia of text styles.

I’m not against good looking documents. I just don’t think there is more value in having the tool get it right than forcing the user to deal with it. We don’t expect users to figure out kerning on their own, do we? They just type “ll” and the system figures out the correct spacing. Over the years we’ve seen more and more of the typography and visual layout occur automatically.

The goal was to determine if they need to spend money for MS Office 2007 or go with OpenOffice. They used MS Office 2003 before that.

There are interesting highlights:
1. External shape is less important then internals.
2. Changes between MS Office 2003 and MS Office 2007 are skin-deep.

First both MS Office 2007 and OpenOffice.org were shown and users were asked to voice their opinions. Most users felt they'll be comfortable with OpenOffice.org (it looks almost like MS Office 2003!) and will be uncomfortable with MS Office 2007 (it does not look like a MS Office at all!).

Then they were asked to perform series of tasks (create letter, different types of reports, etc). After that their opinion was asked again. It was TOTALLY reversed. MS Office 2003 was still the preferred choice (this is what they know best), but MS Office 2007 was deemed "acceptable" and OpenOffice.org "almost unusable".

Why? Here is one task: "you've gotten letter without page numbering, add page numbers at the bottom". Failure rate for OpenOffice.org: 85%. Success rate for MS Office 2007: 85%. How come? Menu items are completely rearranged in MS Office 2007 and moved only slightly in OpenOffice.org (from "Isert Page Numbers" in MS Office 2003 to "Insert->Fields->Page Number" in OpenOffice.org)! The fact that items were moved in strange place affected just 30% of probationers. 70% of users were able to find the item. Yet 3/4 of them were unable to finish the task AFTER finding this item. The difference is simple: if you insert Page Number both MS Office 2003 and MS Office 2007 will ask you if you want this item at the bottom or at the top – and what format you'll need. OpenOffice.org will just put Page Number field where the cursor is. You need to first create footer and then put field there. This idea was insurmountable for most probationers.

That's the biggest problem with your "let's rethink the Office" idea: people are ready to accept cosmetic changes (they accepted MS Office 2007), to teach them new way of thinking… it's much harder…

For example, I can point to languages, such as Egyptian, Sumerian or Akkadian, where expression of semantic categories is directly expressed in the written language. So the Sumerian sign for “LUGAL” will always precede the name of a king, where in hieroglyphics, the name of a god is placed within an ovular cartouce. Although I cannot prove that these cultures did not die out because of their use of semantic tagging made thinking too hard, my guess is such linguistic features are not incompatible with the human brain.

We essentially do this in English as well, when expressing possessives, e.g., “Mike’s dog”. The ‘ tags it as a possessive although this is purely a written convention, and no child learning to speak the language would know about it. But still we semantically tag possessives when we write formally. Similarly we tag proper nouns with capital letters, put foreign phrases in italics, computer keywords in monospace, etc.

The problem is not that we do not know how to think semantics when we write. We do it all the time. The problem is that our tools force us to think about them purely in terms of typography rather than semantic categories. That is the unnatural part.

So say I am writing an article and I want to put in a code fragment. Which is the lazy, natural approach: choosing to indent the block, turn off full justification and set “courier new” as the font? Or simply pick the “code fragment” style? Unless you are a graphic designer, you are probably not thinking in type. You are thinking about the subject matter at hand and the semantic approach is much closer to what your are thinking — lower impedance mismatch — with fewer distractions and context switches than trying to apply text attributes directly.

It is hard to argue for the essential nature of bold and italics. These did not exist until relatively modern times. We were doing semantic tagging for 4,000 years before before italics even existed. I won’t deny that there is convention regarding the use of the bold and italics button in word processors. But it is a 15 year old bad habit only, not a rule of thought.

If they will just get over it, why has semantic formatting consistently been ignored in favour of presentational formatting?

(Quick answer: because thinking clearly is too much work. I’m not saying it’s a good thing, I’m saying it’s an unavoidable design constraint. You can’t design an international standard by starting “First, we need better humans.”)

I’m talking about tendencies to how people use things – not whether it’s a horse or car, but the approach – and why your point 5 shows fundamentally erroneous thinking because it treats the thinking patterns of most people as a bug, rather than an unavoidable design requirement.

I don’t see that they were driving cars with any more attention to what they were doing when they were riding horses.

@David, Before Henry Ford came alone, any survey would just say that all people wanted was a good horse. Maybe it was because of familiarity, maybe some emotional attachment. Ditto for the first electronic spreadsheet, the first electronic word processor, the first wiki, the first web browser. Innovation starts from asking “what if”, not by asking what other features the user wants to see added to the familiar Word interface.

Remember, people once wrote everything in a big cursive hand on pieces of paper with an quill pen that they cut themselves and used ink that they mixed up themselves — ultimate direct manipulation. But they got over it. They once manually controlled line wrapping by deciding when to press the carriage return lever on a typewriter. Can’t more like direct manipulation than that. But we got over it.

All of these innovations involved loss of control. All of these advances looked scary at the time. But in retrospect they are all obvious. Some of them, like the loss of hand-cut pens, led to the demise of entire forms of expression, like calligraphy. But we move on.

@David, if I were designing a word processor for children, I might agree ;-) But seriously, if we design only for people’s inclinations, then we would design a hammer so it stored beer and was soft on the head so you could sleep on it, and was long enough to scratch your back. But a hammer is a tool and we need to design it for the appropriate task — hitting nails — even if that task is not particularly fun. So also a word processor. It is used for preparing documents, and the nature of that task has evolved and we need something that increases productivity for that task. A user’s perception of intuitiveness is 90% by analogy with familiar paradigms. After sufficient exposure to a new paradigm, it will be hard to image going back.

That drives computer scientists up the wall. But it’s also a description of how children pick up language in the first place. So of course people are going to use HTML, wikitext, document creation programs, handwriting, pens, pencils, paper, etc. the same way.

In the context of ODF: how do we think of the tag-soup nature of humans as a *feature*? Because thinking of it as a bug is going to put you fundamentally at odds with the vast majority of the humans you’re supposedly doing this for, and will be a sure sign that your thinking is deeply flawed.

In recent study some secretaries were asked what they are supposed to do: write letter with headings and such. It goes without saying that none of them used styles. This is not what shocked me. The fact that over 80% of them used spaces to center the title and text in the top-right corner did.

If users find one tool that works they’ll use it 9 times out of 10. Even if better tool exist. And the tools which force usage of complex stuff (styles, etc) lose in marketplace.

I don’t know how new format can change it…

HTML is supposed to use semantic formatting. When humans who weren’t nuclear physicists got their hands on it, they went hog-wild for presentation formatting.

One of the biggest things holding MediaWiki back in the enterprise (over *proprietary* rivals such as Confluence) is that (for obscure technical reasons) it’s very hard to do good WYSIWYG in it.

People want and demand direct manipulation of the presentation, no matter how much and how persistently computer scientists keep trying to make them use semantic formatting. Semantic formatting is a fantastically useful and powerful idea and the only way to do things really properly. But most people don’t give a hoot, any more than they’re itching to get off their Windows box and write shell scripts.

Poeple use word processor as mini-DTP to create birthday invitations. People use spreadsheet to keep lists. Both of those use case you would make much more difficult with removing or hiding visial formatting toolbar [buttons] in word processor, and removing grid from spreadsheet.

Re: the artificial dichotomy between spreadsheets and documents —

WordPerfect bought Quattro from Borland to round out their “office” suite; Microsoft bought Excel for the same reason. Neither of them made any effort to merge the two products into a single application. They were both competing with Lotus 1-2-3 and that’s the paradigm people were demanding.

Nearly 20 years later we are still stuck in an artificial paradigm created out of the initial limitations of computing and the resulting narrowness of vision.

Microsoft tried to merge some of this with OLE, but that didn’t really work because you can’t natively refer to the results of an embedded spreadsheet later in your document.

It was actually my early impression of OpenOffice that it had already merged the two, and I was disappointed that they were still separate apps…

It has always struck me as ironic that the Mozilla foundation took the browser in the opposite direction — Mozilla was an integrated application; now they’ve spun off the browser and communications clients on their own (Firefox and Thunderbird) and abandoned the composition client.

#5 – forcing the use of styles — this is what Ventura Publisher did back in the late 1980s and early 1990s. Additionally, the stylesheet was a separate file from the document. This made it a snap to typeset database reports because you could insert the text tags as part of your report, open it in Ventura and the formatting was instantly applied. This is the right idea and the direction documents should have gone long ago. CSS allows the browser to select different layout styles for the same page, but very few sites give the viewer the option to switch styles. Documents should do this.

It is hard to generalize on this. Standards are produced when you get together a bunch of smart people with expertise in a particular area, typically representing vendors. As such they are inherently no more or less likely to produce something of value than any other assemblage of smart people with expertise in a particular area. If anything the trend is for technologies that are by nature network goods start out as multi-vendor standards and commercial success starts only after that. For example, it is hard for a single vendor to push a new optical disc standard, since commercial success requires cooperation and interoperability among a wide range of parties: content providers, publishers, electronics manufacturers, etc. In the case of office document formats, I think have been transitioning to where they are also now represent network goods, and customers increasing demand that their formats are standardized. This will naturally lead to innovations in these formats to also be standardized.

Certainly my blog post is speculative, at least when applied to office formats. But most of these ideas have been applied successfully in other domains. In any case, the ODF TC should at least review speculative ideas in ODF-Next, and decide which ones have sufficient support to go forward with. If this leads us to do something innovative, then that’s a risk I’m willing to take.

I think it is a great list to consider in developing products that have those qualities, and in learning how they might fit into the practices of people with work to do. Some standardization will help, certainly, as interoperability concerns arise, but I don’t think this is something to be solved by speculative standards-writing.

I thought we already had enough failures with using standards writing to invent technologies. Is it just that memories are short?

The only successful exceptions I can think of are projects under the IETF, but they don’t grant the “Standard” imprimatur to anything that hasn’t been built and become well-established.

Ignoring the particular interpretation of how we got here and what is or is not valuable, I take this as a terrific basis for a checklist on what we should be asking ourselves in the development of new ways to look at office productivity software and electronic “documents.”

I am also rereading “Dreaming in Code” word by word and wondering what is that separation between dream and reality that plagues us without apparent end.

There might be something we can learn from Knuth’s Literate Programming as well. The verbal description of a model and the calculations of the model should be closely integrated, so the spreadsheet can be viewed as calculating engine or a description of the model and the results, depending on your mode.

In any case, I think we’ve artificial divorced analytic and textual capabilities. This is probably because 1-2-3 came from one company and WordPerfect came from another. But there is nothing in the nature of the underlying task that suggests these are different.

As for getting people to use styles rather than directly setting text attributes, I think it is a matter of comfort level. You need to show them that the document will consistently look good, even if they do not obsess over it. In the same way, the first time I moved from a manual typewriter to a computer, I felt I lacked control because I did not have a bell to tell me I was near the end of a line. In fact I was told not to explicitly add a carriage return on every line that the word processor would take care of it. It took a few days to unlearn the old habits, but now using automatic line wrapping feels natural and increases productivity.

This is an example of power being what you can ignore rather than what you can control. Heck before typewriters, laying out type required placing strips of lead between lines (so-called “leading”) to control line spacing, and other techniques to handle horizontal spacing (“tracking”) and kerning. Now, very few of us worry or even know about this. The word processor controls it and gives consistent results, better than 99% of us could do if we tried to mess with it manually. Same thing with bold, italics, etc., IMHO. We need to let go and ask ourselves, “What could I be doing of value to my company/business/other endeavor with my time rather than micro-managing the typography of this document?” More and more of the process is automated. This is progress.

BTW: the word processor LyX (based on LaTeX) has some of the desirable characteristics you mentioned, in that the user decides what they want to do, not how they go about doing it.

Mike S

14 – coupla of typo’s. Also sounds aspirational vs. today’s world. The answer to this is, perhaps, as you suggest elsewhere to only allow formatting through styles, but many (most) people don’t actually understand styles.

23, 30 – Seems to me this gets to the need to divorce the subject matter access scheme (e.g. reference coordinates) from the cells and allow those same references to point to objects… Spreadsheet cell references are just a convenient way to represent those underlying object locations.

32 – why should there be a separate “authoring” environment. I’d like one environment for authoring and consuming… editing, formulas, functions, calendars, communications (email, IM, browsing), formatting (why shouldn’t I be able to reformat other peoples incoming data), unstructured note-taking, action lists, searching, structured data, etc.

34 – doesn’t really belong in this list. I agree tho, that MS is the biggest prisoner of their choices. Keep pushing ODF forward for us, I’m finally starting to see the bars come loose from the windows.

Thanks for starting the dialog!

Does a user really need web-enabled, interactive, media drenches word-processing documents? For most use cases, simpler is better.

I agree with the other poster, you must have accidentally drank the Web 2.0 baloney koolaid this morning.

There is a reason Google’s search page is the most popular. Users want simplicity, lack of clutter, lack of all the other crap vendors want to sell to them and confuse their lives with.

The difference between a spreadsheet and a word processor is arbitrary. Other applications have factored analytic and textual functions in entirely different and equally valid ways, e.g., Mathematica, MathCAD or wikiCalc. Personally I think that application-hopping from word processor to spreadsheet and back is an artificial way to work. It was probably invented as a way to market the multitasking features of Windows 2.1.

@Nick, We certainly are seeing the technology market evolve and things that once were high margin are now commodities. RAM, bandwidth, storage, etc., are all cheap now. But software requires highly-skilled labor, and labor prices generally rise, not fall.

To the extent open source upsets the economics of software, its impact is disruptive and revolutionary. I think this is a good thing. If I told you I had a way to feed all the people of the world for free, I’d win the Nobel prize. But if I tell you I can give them all the software they need for free, I’d be called anti-innovation.

@Bart,

Thanks. I’ll take a look at Amaya.