Impact of release of WinFS on computing?

[C. Wyvern]

Realistically speaking, what would computing be like today if WinFS had been released? Would there have been a fundamental paradigm shift in the way computers are used, even now over a decade later with the advent of mobile computing and cloud services/storage?

Or would WinFS have faded away with the advent of such technologies?

[mrpijey]

The tech in WinFS was released, as part of SQL Server. But what you refer to (as a pure file system), no. It would be a file system. NTFS didn't change how computers were used, nor did APFS change how Macs were used. Sure, the underlying layers on how to store metadata and all that would change of course, and it would bring benefits in speed, resilience and security, but the file system doesn't really define how computers are used.

[AlphaBeta]

First of all, WinFS wasn't even supposed to be a file system in the first place. It was a layer on top of NTFS.

The concept of files and individual pieces of information linked together in a database that WinFS conveyed was nice. You could link an event to a bunch of photos and videos, e-mails, any files in fact. Now with stuff like cloud storage that we have today, I reckon WinFS would be even more powerful and effective. On the other side, wouldn't this be a privacy nightmare? After all, it's easier to steal stuff from a tidy room than an untidy one.

In the end, I don't think it would change the way we use computers a lot, because WinFS wasn't even supposed to do that. Its point was to create relations between data, so it could be better managed. I am pretty sure it would increase productivity though.

[TinaMeineKatze]

This is a very good question. Contrary to the others who posted until now, I believe it would have changed how people used computing radically. The idea that data could be linked in such ways would heavily influence the way of working with a computer and creating software for it.

More and more software would interconnect in ways that we cannot even imagine now (just like the internet, technically, nothing so new, but the way of getting information this way radically changed everything in our societies). It would be, a. a miracle for technology to deal with data abstracted away from concrete storage forms (like XML, JPG, ICS, ...) in such a way that more and more connections would be visible that otherwise went unnoticed (like relationships between people that could be interfered from the data, like "I have contact X in my email and he's in 1000 Photos, so he must be somehow important to me"), would largely impact the way data could and would be used. And, as everything in life has it's two sides, it would be a nightmare from a privacy point-of-view. We see how Facebook grew as big just with data, interconnected in many ways, and maybe even influence politics.

Every file whose information you can access is, from a programmer's point-of-view, more work. Even if it's just loading a JPG module, it still needs to be done. As far as I've understood this, WinFS would have changed that, so that every application can access anything in it's store without understanding it's internal structure. And this is really helpful in my eyes. This way, applications can work with a lot more than programming was needed.

Also, I've had a BlackBerry Passport for the last half year, and there, something similiar (though way simpler) is done. I can join like email, name, address, ... with one contact and while in his e-mail, I can just tap once to call the person (since blackberry knows the relationship of the email-address and the phone number). Two more clicks and I can enter his address automatically to the map navigation tool. This, for me, is really a killer-feature of that phone. So expanding this idea way more abstractly would've certainly been a killer feature for Longhorn, at least for me and everybody who's interested in getting-stuff-done.

If Microsoft created WinFS with the attitude of Windows 10, they'd wanted know everything. Nothing would stay a secret, even information that we can't even imagine come together.

There's a nice talk, sadly, only in german about a data scientist who scanned spiegel.de (Germany's biggest news site) a few thousand times a day for new articles and archived them. With that (only the public articles!) he was able to conclude when people had holidays, when they worked, what times they worked, what members constitute what teams internally and who's probably in a love-affair inside spiegel.de's headquarter and with whom. (For everybody able to talk german: https://www.youtube.com/watch?v=-YpwsdRKt8Q ). Imagine what you could do with malicious intents with every personal data accessable easily and Microsoft not caring about privacy all too much.

I would still really enjoy a linux-open-source-implementation of the WinFS idea.

[C. Wyvern]

This is a very good question. Contrary to the others who posted until now, I believe it would have changed how people used computing radically. The idea that data could be linked in such ways would heavily influence the way of working with a computer and creating software for it.

More and more software would interconnect in ways that we cannot even imagine now (just like the internet, technically, nothing so new, but the way of getting information this way radically changed everything in our societies). It would be, a. a miracle for technology to deal with data abstracted away from concrete storage forms (like XML, JPG, ICS, ...) in such a way that more and more connections would be visible that otherwise went unnoticed (like relationships between people that could be interfered from the data, like "I have contact X in my email and he's in 1000 Photos, so he must be somehow important to me"), would largely impact the way data could and would be used. And, as everything in life has it's two sides, it would be a nightmare from a privacy point-of-view. We see how Facebook grew as big just with data, interconnected in many ways, and maybe even influence politics.

Every file whose information you can access is, from a programmer's point-of-view, more work. Even if it's just loading a JPG module, it still needs to be done. As far as I've understood this, WinFS would have changed that, so that every application can access anything in it's store without understanding it's internal structure. And this is really helpful in my eyes. This way, applications can work with a lot more than programming was needed.

Also, I've had a BlackBerry Passport for the last half year, and there, something similiar (though way simpler) is done. I can join like email, name, address, ... with one contact and while in his e-mail, I can just tap once to call the person (since blackberry knows the relationship of the email-address and the phone number). Two more clicks and I can enter his address automatically to the map navigation tool. This, for me, is really a killer-feature of that phone. So expanding this idea way more abstractly would've certainly been a killer feature for Longhorn, at least for me and everybody who's interested in getting-stuff-done.

If Microsoft created WinFS with the attitude of Windows 10, they'd wanted know everything. Nothing would stay a secret, even information that we can't even imagine come together.

There's a nice talk, sadly, only in german about a data scientist who scanned spiegel.de (Germany's biggest news site) a few thousand times a day for new articles and archived them. With that (only the public articles!) he was able to conclude when people had holidays, when they worked, what times they worked, what members constitute what teams internally and who's probably in a love-affair inside spiegel.de's headquarter and with whom. (For everybody able to talk german: https://www.youtube.com/watch?v=-YpwsdRKt8Q ). Imagine what you could do with malicious intents with every personal data accessable easily and Microsoft not caring about privacy all too much.

I would still really enjoy a linux-open-source-implementation of the WinFS idea.

My favorite response thus far.

For Windows 10 you briefly touched upon the telemetry. I mention that Microsoft discussed in a presentation gathering opt-in telemetry over WinFS circa 2003-2004.

Do you think some end-user scenarios have since been realized with the introduction of Windows Search? There is of course no real semantic meaning between the data, no rules or relationships, and no programming model to make Pangaea (the separate silos or 'islands' in which data still reside) whole again, but there was for example an advent on the use of metadata of which WinFS was seemingly so fond (ex: Dynamic Sets). You could now do that today if you wanted to search for, say, genre of music of specific length with certain rating. Many others.

Are there other scenarios where WinFS would have changing computing? It is so strange to see the industry veer in the seemingly opposite direction, that of cloud computing for storage. It does not solve the issues that WinFS was intended to solve. There are no relations. There are no rules. Few actions for users. Limited metadata capabilities in even Microsoft's SkyDrive. The cloud is a dumb blob store.

[TinaMeineKatze]

My favorite response thus far.

For Windows 10 you briefly touched upon the telemetry. I mention that Microsoft discussed in a presentation gathering opt-in telemetry over WinFS circa 2003-2004.

Do you think some end-user scenarios have since been realized with the introduction of Windows Search? There is of course no real semantic meaning between the data, no rules or relationships, and no programming model to make Pangaea (the separate silos or 'islands' in which data still reside) whole again, but there was for example an advent on the use of metadata of which WinFS was seemingly so fond (ex: Dynamic Sets). You could now do that today if you wanted to search for, say, genre of music of specific length with certain rating. Many others.

Are there other scenarios where WinFS would have changing computing? It is so strange to see the industry veer in the seemingly opposite direction, that of cloud computing for storage. It does not solve the issues that WinFS was intended to solve. There are no relations. There are no rules. Few actions for users. Limited metadata capabilities in even Microsoft's SkyDrive. The cloud is a dumb blob store.

Thanks.

I didn't know about the telemetry stuff being discussed that early, but it makes sense from a capitalist point of view. Microsoft doesn't really care about customer satisfaction as a big goal. It's just a intermediate goal for making more money (which I want to say without judging that; it's sadly the way companys just work).

And to your question: sure, the way Windows Search works now is a bit more convenient, but still far, far away from what WinFS could have offered. The information linking is what would have made it so powerful, and there's barely anything like that in the computer world now.

I work as a programmer in a research team developing neural networks and cloud solutions, and from my point of view, "cloud" is just a fancy way of saying "we don't care about it, let somebody else do the work", and so, yes, it's just a dumb blob store out of my control technically. Sadly, since most people don't care about doing real work at just like to rely on somebody else to do it, which really waters down the technical abilities.

Maybe some day, there'll be solutions like WinFS for this situation where the data is not stored as a pure blob on somebody elses machine (though I highly doubt it, at least for commercial software, since the information itself is what's being sold and not customer satisfaction or ease-of-use). But right now, I cannot see anything similiar.

As I've said, I believe a real interconnected data store would heavily impact the way we use our computers, we think about computers and how computers shape the everyday life of people (in every area, academics as much as grandma's computer). I remember Tim Berners Lee, the inventor of the internet, saying in an early paper about the net that it allows people to "get information they didn't even knew they needed". Any WinFS-idea-like-system would probably do the same. But sadly, I cannot look into alternate realities where WinFS was fully developed and used, so I have no clue how exactly this would impact the world of computing.

I can only really look forward to a free (as in free speech) implementation of this idea.

Would anyone know of such a thing? I've never heard of something similiar.

[jimmsta]

If anything, Microsoft has re-iterated on all the attempts of getting a true Object Filesystem into NT since the early beginnings. WinFS was a bit different, as it added a huge amount of metadata linking to the Object idea. If anything, the ObjectFS idea was implemented in the various Stores that NT Server utilizes for things like Exchange and Active Directory. Both of those products require large amounts of linking of objects to file data. It's not all dead in the end. WinFS tech, being from the .NET era re-iterated the OFS ideas, and aside from being integrated into SQL Server, was likely the basis for later versions of Exchange and Active Directory's data stores.

In very specific use cases, as outlined above, this tech thrives. It's not exactly exposed to the end user like it started out life as, but it is likely still hidden under the surface.

[C. Wyvern]

I didn't know about the telemetry stuff being discussed that early, but it makes sense from a capitalist point of view. Microsoft doesn't really care about customer satisfaction as a big goal. It's just a intermediate goal for making more money (which I want to say without judging that; it's sadly the way companys just work).

Yes there is (unfortunately or fortunately depending on the perspective). Please consult the supplementary material 'System.Search, aka 'Find My Stuff'' by Chris McConnell, which mentions under Platform Privacy 'Data collection only with [activation of] opt-in'.

And to your question: sure, the way Windows Search works now is a bit more convenient,

Forgive me but this sounds so flippant! I think it was such a big advance from before but apparently much unappreciated. It does solve or help to solve several problems the other solution was intended for.

but still far, far away from what WinFS could have offered. The information linking is what would have made it so powerful, and there's barely anything like that in the computer world now.

I am not sure for second emphasis but...I apologize again. It was not the intention to conflate the two technologies (Windows Search, WinFS) though there are obviously benefits from both. A different supplementary material mentions queries with WinFS [with Dynamic Sets] that are now possible with Windows Search: 'All recently created Word documents', 'All Pictures taken in 2003 of Joe', 'All Rock music that I've rated 5 stars'.

Disregard this! Information Agent in the latter technology is alone with the price of admission. Not even Windows 10 with its so-called majesty and splendor as the latest and greatest has this. No rules to act on notifications based on words or emails or to individually dismiss sounds. The Focus Assist could potentially blossom into providing similar as these, but...no relationships. No data store. No way to relate images taken during a business event in a certain time with events in a user's calendar. Even no rules to act on file management (ex: photo taken before date? move to folder)...

I work as a programmer in a research team developing neural networks and cloud solutions, and from my point of view, "cloud" is just a fancy way of saying "we don't care about it, let somebody else do the work", and so, yes, it's just a dumb blob store out of my control technically. Sadly, since most people don't care about doing real work at just like to rely on somebody else to do it, which really waters down the technical abilities.

Barring viewpoint on cloud, would you consider yourself fortunate in such industry?

Maybe some day, there'll be solutions like WinFS for this situation where the data is not stored as a pure blob on somebody elses machine (though I highly doubt it, at least for commercial software, since the information itself is what's being sold and not customer satisfaction or ease-of-use). But right now, I cannot see anything similiar.

Sadly it seems world has moved beyond such notions. I cannot even get emphasis on metadata in cloud clients.

As I've said, I believe a real interconnected data store would heavily impact the way we use our computers, we think about computers and how computers shape the everyday life of people (in every area, academics as much as grandma's computer). I remember Tim Berners Lee, the inventor of the internet, saying in an early paper about the net that it allows people to "get information they didn't even knew they needed". Any WinFS-idea-like-system would probably do the same. But sadly, I cannot look into alternate realities where WinFS was fully developed and used, so I have no clue how exactly this would impact the world of computing.

Thank you!

I can only really look forward to a free (as in free speech) implementation of this idea.

I know this is not what you want...but it is sad that Microsoft would not try this again. It could undoubtedly be done with the right resources and time, plus with benefit of unfortunate history... Perhaps not possible for many to imagine implications of such powerful technology. I wish there were more answers.

Maybe I should try a different question for others [and to read another one of your thoughts if I am so fortunate]?
What would Windows itself be like in 2018 if WinFS had been released in 2006?

[Maza]

I am not sure for second emphasis but...I apologize again. It was not the intention to conflate the two technologies (Windows Search, WinFS) though there are obviously benefits from both. A different supplementary material mentions queries with WinFS [with Dynamic Sets] that are now possible with Windows Search: 'All recently created Word documents', 'All Pictures taken in 2003 of Joe', 'All Rock music that I've rated 5 stars'.

You are wasting your time. Rather than appreciate what is available most users—I am not referring to any user who posted here—would rather bask in the glory of some mythological era or in the shadow of some technology that will unfortunately never see the light of day.

Just look at all of the YouTube videos for "WinFS" where the people can only copy or, ahem, paraphrase the information from Wikipedia and pretend like it is some grand revelation.

Forget that Windows Search objectively provides several intended features.

Forget that even after the development reset there were additional features under the "WinFS" umbrella such as Metadata Painting.

Forget that the feature intended to replicate data across PCs on the same subnet ("Castle") was still included after the "development reset."

Forget that Static Lists were also included after the "development reset."

[SistemaRayoXP]

You are right, some of the intended features were actually developed further. But they weren't developed nor targeted as they were originally meant to be. WinFS ended up in Server side, not consumer, so it benefits only to server managers,not to the actual users.

Another technology is the Windows Search. I will ask you, how many people do actually use any advanced features of the Windows Search? It's a ridiculously little amount of people, because those features do not provide the same functionality, also they are part of the Windows Search, which isn't instant search (in WFS search would be almost instant), so it takes time, and you might even not find the results, what results in time spend, among many other reasons why people doesn't use Windows Advanced Search.

Also, in WinFS all those technologies were meant to work together, so they can be more efficient. But they ended benefiting different parts of the OS, so they are not as efficient as they were originally meant

[Maza]

You are right, some of the intended features were actually developed further.

This is an unappreciated detail.

But they weren't developed nor targeted as they were originally meant to be.

For the end-user experience this means so very little if anything at all. I would rather have "WinFS" of course because of its other benefits, but I am not going to be actively realizing what the underlying technology is being used when I, for example, stack my items over metadata properties—a key "WinFS" differentiator that was realized with Windows Search in Windows Vista.

I am obviously not intending to conflate the two like the OP seemed to be but, you know, on that note, the stacks feature by itself actually did have advertised "WinFS" features even after the "Longhorn" debacle, in Windows Vista Beta, before Microsoft decided to gradually pull the features because testers were confused with all of the changes that the new Windows Search vision entailed. Users had the dragging and dropping items into stacks to write new—or modify existing—metadata properties. Users had the custom thumbnails for stacks. Users could write properties to stacks themselves to affect all of their referenced items! Users could rate an Album stack, for instance, and this would assign that rating to all music tracks referenced by the album (in a manner similar to Windows Media Player).

For other features such as static lists and virtual folders (search folders) this went further. Dragging and dropping a virtual folder into another Virtual Folder, for instance, was intended to write the metadata properties of the latter to the items referenced by the former, as shown in the following table from Windows Vista Beta 2:

Another technology is the Windows Search. I will ask you, how many people do actually use any advanced features of the Windows Search? It's a ridiculously little amount of people

I am one of the few. I will tell you that it is literally their loss if they do not use the advanced features—it is tragic, and it is a shame.

because those features do not provide the same functionality

How can you write that this is the reason for the advanced features not being used? The vast majority of end users have not even heard of the advanced features let alone "WinFS." The users who do not use the advanced features are the same users who have never had a taste of true sensation. There is accordingly no incentive to discover or use such features.

I think you might be surprised to discover that several of the end-user scenarios discussed in the PDC 2003 "WinFS" AERO Guidelines are actually possible now—today!—only because of Windows Search. Here are screenshots that compare the stacking features from "pre-reset" "Longhorn" with "WinFS" (left) and "post-reset" Windows Vista with Windows Search (right). The same guidelines state that "With 'WinFS' users can access anything they need by familiar contexts rather than by digging through a hierarchical folder tree." With Windows Search I do not have to care—indeed, I do not need to know—where an item is stored in some buried folder structure.


It certainly does not help that since the release of Windows 7 Microsoft has gradually shifted the focus away from such end-user scenarios that rely on Windows Search. Libraries are not worthy of their name—even the "Library" virtual folder of Windows Vista Beta 2 was superior.

Here are some of the other scenarios possible with Windows Search:

• Stack all audio by the Album property and filter to display only the albums with the tracks that are rated 5 stars;
  Stack all audio by the Track Number property to listen to each song with a certain track number (such as the very first song from every album);
  Or more advanced queries like all email sent between 1990-1999 that is from a specific address and that has attachments;
  In Windows Vista and Windows 7, the natural language search feature allows users to literally type "email received last year from contact about subject"

An amazing thing about these queries is not just that they can be saved for future use but that they can also be shared with other users.

Why tell you about the search folders? When I can share them with you?!?

also they are part of the Windows Search, which isn't instant search (in WFS search would be almost instant) so it takes time, and you might even not find the results, what results in time spend

Windows Search is instant for me. "WinFS" also relied on content indexing.

Also, in WinFS all those technologies were meant to work together, so they can be more efficient. But they ended benefiting different parts of the OS, so they are not as efficient as they were originally meant

The extent of this intended unification was of course better, but Castle, for example, was designed to work in tandem with Windows Search, returning results of files from other PCs. They were not completely separate components.

[SistemaRayoXP]

This is an unappreciated detail.

If it is so, why do you comment it?

I don't want to disarm every single of your arguments, I don't want to make a really large post, but resumed, if I understood correctly, you are saying that most of the features announced in the PDC 2003 ended up in the final Vista, and you have shown proof of that. But you are forgetting one thing, the average user.

By what I'm reading I understand that you have tweaked up your system adjusting it to your needs, but the average user who doesn't know how to index the content, for example, will never be aware of this, and even I personally have never indexed my content (Reason doesn't matter now). The main difference is that you have "dominated" the Windows Search and the metadata. But any average user wouldn't do this because ignorance respect to how to use these features. Another reason might be that MS never really emphasized these features after the reset, and so the end user was never aware of this features, and MS never taught average users about how to use the features, so it's really down to the user to inform.

At the end, only the advanced user will benefit from the Windows Search, when (AFAIK) WinFS was announced as a main feature, something that the average user would care about, not like the Windows Search, which MS placed at less relevance for the final user 8Again, unlike WinFS)

[Maza]

If it is so, why do you comment it?

Another very excellent question from you! Please recall that I wrote that it is tragic that other users do not use these features. My primary motivation is to glorify Windows Vista and to inform those who are unaware of these features.

I don't want to disarm every single of your arguments, I don't want to make a really large post,

Understandable. Thank you.

but resumed, if I understood correctly, you are saying that most of the features announced in the PDC 2003 ended up in the final Vista, and you have shown proof of that.

Those actually demonstrated? I would say "perhaps yes" ("post-reset" Windows Vista beta actually had more)! However, I am very careful here and I realize that Windows Search is not "WinFS." The OP already mentioned Information Agent, for instance, and rightfully stated that there is no equivalent today. It is difficult to fathom the power of what are effectively Microsoft Outlook Rules and Alerts across entire items, or relationships between items without necessarily relying on the underlying metadata, or commonality among previously separate data types (e.g., contacts).

But you are forgetting one thing, the average user.

The average user is of course the most important, largely because such users are so numerous.

I do not mean to forget them, as it would seem, and I intend to address such shortcomings in the future with the release of many virtual folders that work over metadata properties like those that work in Windows Vista Beta or, in several cases, "pre-reset 'Longhorn.'"

By what I'm reading I understand that you have tweaked up your system adjusting it to your needs,

Somewhat. Only with the stacks and the virtual folders that I use literally on a daily basis—the rest is handled by the system by default.

but the average user who doesn't know how to index the content, for example, will never be aware of this,

That is a good point; however, content is indexed by default in Windows starting with Windows Vista. This should not be an issue for the average user.

and even I personally have never indexed my content (Reason doesn't matter now).

Ever? Are you sure (in light of the above)? I do not mean to question this or your honesty, but this is difficult to believe!

The main difference is that you have "dominated" the Windows Search and the metadata.

Well ah, I really like that term that you used. Thank you so much.

But any average user wouldn't do this because ignorance respect to how to use these features.

Of course. I blame Microsoft for this (for stripping away metadata painting, for instance, which would provide a rich experience to users to write single or several metadata properties to items without ever them even touching the keyboard). The majority of the blame is due to the treatment of the Search and Organize vision with the release of Windows 7 which. For example, the operating system removed the default virtual folders from Windows Vista: Recent Documents, Recent E-mail, Recent Pictures and Videos, Recently Changed, and Shared By Me (it is interesting to see that Microsoft still grouped pictures and videos by default with the search technology in Windows Vista, as per earlier "Longhorn" concepts).

Another reason might be that MS never really emphasized these features after the reset, and so the end user was never aware of this features, and MS never taught average users about how to use the features, so it's really down to the user to inform.

I most agree, as per above. Yes.

At the end, only the advanced user will benefit from the Windows Search, when (AFAIK) WinFS was announced as a main feature, something that the average user would care about, not like the Windows Search, which MS placed at less relevance for the final user 8Again, unlike WinFS)

I think I now understand what you meant before. I interpreted your previous comment as meaning that Windows Search would not be beneficial or at least not as beneficial.