Well-Known Apple Developers Support Manifesto for Ubiquitous Linking

I could be wrong, but IIRC, OpenDoc was for Macs only, and it was document focused. Developers were turning away in droves from Macs, and they had ramped up full stream ahead and were cranking out stuff for Windows. And MS Office had already pretty much conquered the world. Adobe had stated it would stop updating Photoshop and its other apps for Macs, and other companies were doing so as well.

At the time, Apple was more than just broke, it was heavily in the red. Steve Jobs, who had just returned to Apple, killed OpenDoc to focus Macs on the recently introduced WWW, communications and multimedia as well as documents that were sitting on your hard drive. The very super cool first version of the iMac and OS X were key to this strategy. Steve repositioned Mac OS as a platform that developers would want to build apps for, and one that consumers would have a lot of options to pick and choose from. He focused on what people would want and need.

(My favorite systems support what Iā€™d call ā€œone-hit searches,ā€ where your search switches to or displays the desired item, without an intermediate results list. LaunchBar works that way, as did Googleā€™s now-defunct Browse By Name feature;

A few months ago I realized that while I thought I had my Firefox searches configured to use Google Browse By Name, actually Mozilla had killed it some time ago without notice. So I used a different method to re-enable it.

What Iā€™m seeing now is the quality of BBN results has gone way down: it frequently takes me to the wrong page.

It used to be that BBN would only go directly to a web page if Google was reasonably sure that it was the right, definitive result. But now it is tending to go directly to a page that is clearly the wrong result, or a case where no result should be definitive.

For example: the BBN result for ā€œsalmonā€ is Garlic Butter Baked Salmon - Cafe Delites. What, a random recipe for a particular salmon dish is the definitive page for salmon? Tell me Iā€™m not crazy.

Another example: the BBN result for ā€œrefrigeratorā€ is https://www.homedepot.com/b/Appliances-Refrigerators/N-5yc1vZc3pi. Itā€™s like Google is just taking the current top search hit as the ā€œdefinitiveā€ BBN result. Or is it to whatever site has bid highest for an ad keyword?

I wasnā€™t aware it was still working at allā€”what method did you use to reenable it?

Change the BBN URL to http://www.google.com/search?ie=UTF-8&sourceid=navclient&gfns=1&q=query. (i.e., add the ie=UTF-8 parm).

But then the problem in Firefox is that you can no longer change the search URL it uses except by installing a search engine defintion. So I searched for browse by name at Mycroft Mycroft Project: Browse By Name Search Engine Plugins - Firefox IE Chrome, clicked on the link for Google US (Browse By Name), and then while on that page you can add it as a search engine in Firefox.

Add the search engine by clicking on the URL so it drops down suggestions. At the bottom is ā€œThis time, search with:ā€. 4th from the right will be an icon to add the search engine for the current Mycroft page.

Another problem I found is that the URLs for Google Iā€™m Feeling Lucky searches no longer work; instead of going directly to the site, they stop at a landing page. I think this was because IFL was being abused. The work around here is to use Iā€™m Feeling Ducky instead: https://duckduckgo.com/?q=!ducky+term.

Note: Now Iā€™m really thinking that BBN has devolved into Browse By Ad Keyword.

I was able to install the plugin you mentioned into Firefox, and I could show that it worked by searching for ā€œAppleIDā€ and for ā€œJamf Now pricing,ā€ both of which went exactly to the correct page I wanted. (Lots of other searches didnā€™t, including ā€œTidBITS ubiquitous linkingā€.)

However, I canā€™t make the URL you gave work in Brave or Chrome. Everything just results in a Google search results page. Is that what you were saying when you said that Googleā€™s Iā€™m Feeling Lucky no longer works? (But then how does the Firefox plugin work?) The ā€œTidBITS ubiquitous linkingā€ search did work in Iā€™m Feeling Ducky.

BBN and IFL are supposed to do different things. Iā€™m Feeling Lucky always goes to the first hit in the Google Search results. Browse By Name is only supposed to go to a result if what youā€™re searching for is a ā€œkeywordā€. So BBN for ā€œTidBITSā€ should go to the TidBITS site but ā€œTidBITS ubiquitous linkingā€ isnā€™t a keyword, so it shouldnā€™t go directly to a site. IFL (and Iā€™m Feeling Ducky" would go there because it is the top search result.)

(This is because from what I remember, BBN originated as ā€œkeyword searchā€.)

What I meant about Iā€™m Feeling Lucky is that the IFL URL https://www.google.com/search?q=term&btnI now stops at a Redirect Notice.

The BBN name URL does work for me (as Browse by Adword) in Chrome. For example, http://www.google.com/search?ie=UTF-8&sourceid=navclient&gfns=1&q=ford goes directly to the Ford page. But it doesnā€™t work in Safari. I think Safari is cheating, and is rewriting the search as a standard Google search for some reason. Maybe it has something to do with whatever default search engine contract Apple has with Google.

Interesting! That link does work for me in Firefox and Chrome and Edge, but not in Brave or Safari. So yeah, there must be some behind-the-scenes shenanigans going on.

It does work in Safari with the Safari Keyword Search extension.

http://safarikeywordsearch.aurlien.net

Our lives are full of links. We keep most of them in our head, which is less than idea. As David Allen of GTD fame says ā€œYour mind is for having ideas, not holding them.ā€

There is a Tidbits article that has stuck with me through the years where a physician described how he used a Newton to manage his life. Much of the article described how he could carry around reference material (better today) and manage his scut list (to-dos). Buried in the middle of the article is a section titled ā€œLink It All Togetherā€ where he describes how he can create a note with links to Contacts, Appointments, lists, etc. ā€œItā€™s remarkably easy to create all these links.ā€ It seems that in 1998 the Newton could do something I canā€™t do with the Apple apps on my iPhone today.

Thanks for the article. I signed the Manifesto.

3 Likes

Apple has some private linking capabilities, too: if you click a date in a Messages conversation and create an event in Calendar, it will have a link that opens the original item in Messages.

Unfortunately, this has been broken for a while and remains broken in Monterey.

It used to be in macOS that you could actually drag an email (say with the invitation for a meeting or containing its agenda) onto a calendar eventā€™s info box towards the bottom where it says ā€˜Add Notes, URLs, or Attachmentsā€™ and from then on that calendar item would display a clickable link to that email. This was incredibly useful because not everybody creates calendar items by clicking on the date/time in the email (as the quote above details) they want to link to. At my work, often an invitation will come as a generic Calendar invite, but lots of follow up technical details will be emailed around by other people in other email chains. This calendar linking functionality even held the promise of allowing several emails to be linked to one calendar item because some of us are used to having many emails with technical details related to a single meeting.

Anyway, unfortunately, this functionality only rarely works these days. Iā€™m not sure if itā€™s related to true IMAP vs. bastardized Gmail IMAP or if itā€™s related to shared calendars (Gmail again) vs. local calendar on your Mac, but it clearly no longer works consistently. Even if you succeed in getting Mail to create the link, not rarely will later clicking it just dump you into a Safari window where the ā€œURLā€ appears to be the text describing the link in the calendar entry. WTH??? To add insult to injury, even when it works it breaks if you then for example use iOS Calendar and tap on the link from there, regardless of the fact that the mail message you linked to lives on a remote IMAP server that both the iPhone and Mac have equal access to. Not even ensuring that the iOS Mail app just downloaded that message and has it displayed will help. Itā€™s just flat out broken with no indication to the user that it cannot be used that way. Quite the opposite in fact, the link being clearly shown as a link just begging to be clicked/tapped.

If this thing worked the way it should this would be a) incredibly convenient and b) nicely leverage Appleā€™s ecosystem where one workflow on one Apple device and OS just continues to work seamlessly on another Apple device running another Apple OS. Alas, weā€™re far from that here. Meanwhile Iā€™m manually adding mail receipt time stamps like 20211210-1456 as ASCII text to my calendar entries so at least I know where to manually find relevant emails required in that upcoming meeting. Lots of silly manual work. Almost makes it feel like itā€™s 1982 all over again. :rofl:

2 Likes

Thatā€™s @ronā€™s article! A blast from the pastā€¦

2 Likes

The questions that come into my mind here are many and variedā€¦

  • Should links work inter-OSā€™s/platforms, i.e. not just Apple OSā€™sā€¦or would that be too difficult to implement?

  • Who would be best to implement such ubiquitous linking? (Apple, Google, third-partiesā€¦?)

  • How would the code that does the linking be maintained/expanded upon?

  • Would using it be easy enough for most average users, without getting lost in code syntax?

  • Would yet more file formats be required, or could current ones be used?

  • How would personal/private linking be controllably separated from wider linking outside (e.g. web, et al.)?

ā€¦and a heck of a load more!

Really, if such linking were to become truly ubiquitous, youā€™d need an independent body to conceive it and maintain/expand it. Like W3C does for the web, or other trade organisations like (off the top of my head) the current Matter and Thread ones for IoT devices.

Why did Project Xanadu never get off the ground? Presumably a large factor was awful timing, due to the web idea becoming more realised at around the same time, so people were busy focussing en masse with that, rather than the bigger and much more difficult Xanadu ideas. Thatā€™s life for many great ideas, unfortunately.
But also, mainstream commerciality happened fairly soon for the web later on, and that made it a truly viable entity we all count on today.

Relying on one company to run the show would only end-up in a half-hearted attempt, likely resulting in failure, as it wouldnā€™t have the interoperability of something like the web.

The trouble is, the web and IoT have very clear and obvious large scale commercial uses, so companies are willing to throw vast amounts of money at the maintaining organisations for them to get to work, R+D, maintain, expand, iterate.

While the usefulness of ubiquitous linking would be massive on both personal, professional, and societal levels, is there enough commercial leverage to the idea, to gain traction beyond mainly academia? Thatā€™s something I struggle with. And without that, I canā€™t see it happening anytime soon, unfortunately.

The only commercial aspect to the Xanadu idea was micro-payments for linking to parts of commercial works (be they academic papers, video content, photos, or whatever). But again, would this be appealing to a wide enough audience? That would obviously depend on many factors in its implementation, but again I struggle with this, as micro-payments have yet to take off on other projects out there.


(Matter and Thread info here.)

Thread is a low power, secure and Internet-based mesh networking technology for home and commercial IoT products.

Matter is a unifying, IP-based connectivity protocol built on proven technologies, helping you connect to and build reliable, secure IoT ecosystems.

1 Like

I donā€™t see why not. If an app works on Windows and macOS and has a sync mechanism, why in principle wouldnā€™t the links to the same data work on different platforms? Apps providing and serving links is not very complicated at all. Apps have an underlying ID for objects they sync. Thatā€™s what allows you to be able to modify the object on one device and see the change on the other; and when you delete the object from one device, it gets (or normally should get) deleted across devices. (True, if the developers are not very skilled or projects not well managed, there can be problems. MobileMe was a mess for instance.)

Who would be best to implement such ubiquitous linking? (Apple, Google, third-partiesā€¦?)

for object based apps, each developer just needs to do their part.

Files require a bit more work. But cloud services like Dropbox could help, for instance, by providing an API to get IDs and managing deletion and renaming reasonably.

I donā€™t see the need for that at all.

The protocol is a start. app links are served by the local app. Apps already must have mechanisms to gate who can access what data, irrespective of linking.

The manifesto was proposed in the spirit that this is not really required. No standard, no specific API is required; general principles suffice, as laid out in the manifesto.

Compare the agile manifesto. There are varying degrees of agility in software development, but its mission has been accomplished: weā€™ve (devs) have all been affected by agile/lean. Not a perfect analogy, obviously. But it goes to show that change is possible.

Back to specifics: the most fundamental idea here is to provide Copy Link in UI and via automation. Thereā€™s a bit more to it than that (for object creation for instance), but thatā€™s the key idea.

centralization was a problem. We are not suggesting any centralization or standards.

Iā€™ve worked in areas where standards were required. (At Abatis Systems corp. [acquired by Redback Networks], I led for a while the element management group of a multi-service router. It implemented a huge number of standards (TCP/IP RFCs and layer 2 stuff), many very complex (which I needed to wrap my head around). That was required. I also worked in ed psychology/tech where standards like SCORM were proposed, not a success story.) What weā€™re dealing with here is not comparable to either.

Time will tell. On macOS at least there are already a great many apps whose data are linkable in our sense (UI and automation). Hereā€™s a list we at CogSci Apps maintain: Linkable Mac Apps ā€“ Hook.

Even s/w developers are distributed along the technology adoption life cycle. As consumers vote with their dollars, the laggard developers will eventually follow suite. For instance compare:

Compare for instance this tweet Andreas Busch works from home :eu::de::uk: on Twitter: ā€œUnfortunately thatā€™s still not possible with my preferred #pdf app, namely #PDFExpert from @Readdle. Hey @Readdle, can you let me know whether and if so when you plan to fully support @HookProductvT? By the way, merry :christmas_tree::santa:!ā€ / Twitter. There are however several PDF apps (like Skim, PDFpenPro and Adobe products) that support both file level links and deep links (latter via automation).

The list of signatories has been growing ever since the manifestoā€™s publication. And the manifesto is being discussed (Media Highlights ā€“ Linking Manifesto).

Iā€™m optimistic.

1 Like

LucCogZest, really great further info, thanks. Especially for those of us (non-coders) previous not familiar with the topic, here. :wink:

So in a nutshell, this initiative for devs and users is basically (sorry, I love to bullet point!):

  • Getting the Copy Link functionality that already exists for files (objects) to be copyable within apps. Implementation is decentralised, so anyone can already use them.

  • Apps already exist using said functionality, so they have a proven reliability and robust track record.

  • Hope natural technology adoption life cycle push/pull factors, will help influence devs to implement it and thus drive user uptake.

While links to the object itself is a great start, as the original Xanadu mentioned, wouldnā€™t more be needed to deep link to items within objects, and then dealing with the issues of items with commercial value (via payments of some kind)?
Presumably, these are points for later consideration, rather than the immediate need to get basic linking off the ground and truly ubiquitous.

(EDIT: Hook app does deep linking on some level already, so presumably others could already follow.)

Anyway, I really hope this succeeds, as itā€™s sorely needed and something even average users could make use of for basic uses, never mind those doing research work et al.

2 Likes

Bingo. My guess at the time, and still is, that Xanadu focusing on micropayments was not a smart idea by any stretch of the imagination. AOL quickly dominated the early Web. Google was hovering in the background, beginning to crawl every inch of the WWW, as were many other ad networks. And the idea that every document would be preserved, searchable, and everything would be linkable for time immemorial was not realistic, financially or technically. It didnā€™t address the potential of multimedia online. And there was just about no focus on security or safety. Xanadu might sound compelling and enticing, but it just could not work in the real world. Even what was then unconquerable AOL has just about ended up like Ozymandias.

1 Like

Itā€™s worth remembering that the ideas behind Xanadu predate AOL and the Web and Google by many years. My suspicion is that we simply didnā€™t have the processing power (and possibly storage) necessary for something like Xanadu for quite a few years after even the early online services (CompuServe, GEnie, Delphi, BIX, AOL) or the early Web. Xanadu core concepts are probably much more technically possible now (though there would still be scaling issues) but the network effect has ensured that weā€™re too far down the path with the basic Web to ever implement things like bidirectional links and micropayments at a low-enough level to meet Xanaduā€™s design goals.

3 Likes

Quite a lot of years. According to Wikipedia (and what I remember reading years ago), the concept got started in the 60ā€™s, before we even had a modern concept of networking, let alone an Internet or Web. The first proof-of-concept implementation was in 1972 - still long before modern networking.

For reference, Ethernet got started in 1973 at Xerox and became the dominant LAN technology in the 90ā€™s. The first papers that would become TCP/IP were written in 1974, with the first IP standard (RFC 791) published in 1981.

While it could be implemented today, as you point out, it would be very difficult for it to become popular enough to realize its original goals.

2 Likes

Iā€™ve always thought that a lot of insanely great ideas got started at Xerox, but Xerox never quite realized how important and valuable these ideas and technologies were. Jobs and Woz did, and Iā€™ll bet that there are a lot of people at Xerox who were, and are, dumbfounded that they never imagined the size and depth of the consumer and business markets for personal computers and networking. And how important stuff like the mouse and GUI could be.

And IIRC, IBM invested a very significant amount of cash in Apple to help get it off the ground, for which they received a very significant amount of stock. They sold it all on the day Apple went public. I canā€™t imagine how Apple would have evolved if IBM held on to the stock and maybe had a seat on Appleā€™s board.

1 Like

I think techies will look back at the era where apps were in silos as an oddity too, where the importance of navigating between data in different apps was not recognized by the OS vendors, and few consumers realized they needed this ability to (along with the other things in the manifesto). As more software developers get on board, and a snowball of consumers start requiring linkability, the pressure will be too strong for the laggards to ignore. (Microsoft Outlook still does not have an API for grabbing links, for instance, but Apple Mail , Airmail and others do. And Outlook email URLs are a joke, they donā€™t work across OS instances even for same individual. Readdleā€™s Spark is also a joke in this respect. Their URLs do not expose the RFC-compliant email ID. I guess they want to lock their users in.)

2 Likes

Add Macā€™s Open Doc and Cyberdog disasters into the list.