I have been a temporary ex-Tinderbox user a number of times in the past as well; but I always come back to it and every time I do I am glad for it.
Harvesting difficulties: I really don’t think that is the core goal of the application. It isn’t meant to be something like DEVONthink or EagleFiler. It’s a tool for turning ideas into concrete conceptualisations or implementations. I think it can be used as a research tool, and be quite effective at it, but to me it seems more in line with the philosophy of acting as a concept layer above the actual material layer. Each note can be linked to a file on the drive—and once it is linked it extremely easy to retrieve the file and work with it alongside Tinderbox. The information that is necessary to have expressed in Tb can be copy and pasted into it and become a part of the network. But even in a really beneficial document, there is rarely a need for 100% of it to be available from within the concept network.
In my own experience, encouraging bulk document import tends to lead to data bloat. You get huge databases full of 60% or less useful data. I’d rather just select precisely what needs to be known and link to the file for all the rest. Keep the kernel clean.
As for the actual mechanics of getting data in: I really don’t see what is wrong with copy and paste or drag and drop? These methods are very efficient? How is a clipping service any more efficient? It’s just a matter of what you are used to, I think. Services are non-targetable, due to their nature, and thus require a sorting phase after they have been clipped. Paste is direct to context no sorting required. And if you don’t want to think about it you can always just paste your clips into a holding area. Since Tb creates notes dynamically when you paste into the outline—this is extremely simple and easy to do.
I’ve certainly tried clip-centric programs in the past, but honestly I never got it. Most of the time you have to access the service in some menu with the mouse (or contrive some incredibly arcane universal keyboard shortcut that will not conflict with anything), and unless you spend time keeping that menu clean, it can get really messy loaded up with a lot of stuff that has nothing to do with what you need at the moment. Cmd-V is always there, always doing exactly what you expect.
Difference of habit, I guess.
Well, I’m a geek, so the bit about agents doesn’t bother me. There is a graphical interface for that, by the way, and what it doesn’t address, it does show you how to expand by demonstrating the syntax as you build it with the standard drop-downs. But yes, it is pretty geeky once you get past the basics. I will definitely concede that. But I think it is a valid question whether or not that advanced stuff is necessary all of the time? I have plenty of Tb documents that are extremely simple—mostly just leveraging the map+outline+linking features that hardly anything else does. It’s nice to know that if I need to do something I can, though.
I have to completely disagree with the go-Cocoa+Database analysis, though. I wouldn’t mind the interface getting optimised for say, Intel. That’s my biggest grief with it right now. It runs a bit slow through Rosetta on my MacBook; the fan comes on a lot. But it already does most of the things that Cocoa can provide. What were you thinking of that it is missing? Carbon can be a valid alternative in some situations. As for going database? Ugh. No. I love that it is an XML file. I take advantage of that frequently. That was one of the big selling points, for me. I have saved many weeks of manual labour simply by having access to the raw data in a logical XML manner.