Anyone active in using this tool will have noticed that it has not been working for the past week and a half, and as I’m off interstate for a fortnight soon I’d better explain why.
Tumble tree gets its information by a technique called ‘page scraping’, which gathers information by reading the HTML of a page, requesting their servers for more notes, reading the HTML again. It’s not considered to be a good thing to do, but as the Tumblr API doesn’t offer a method to get all notes for a post it was my only option. Which if you are a single person, Tumblr probably don’t mind so much, your impact would be negligable. But involving more users makes the situation far worse.
I knew it wasn’t a great way of doing things, but then found out that their Terms of Service explicitly forbid scraping, which is grounds for closing off Tumblr accounts and potential legal action:
And as this tool gives a growing number of people the ability to make literally hundreds more requests for notes than they might otherwise (not that most people would), I find myself in the position of facilitating a great deal of strain on Tumblr’s already stretched servers. So until I hear back from Tumblr / discuss reasonable limits / get permission to run this tool, it’ll have to remain offline. Some time ago Missing-e users might remember the warnings that Tumblr put up regarding that extension - and page scraping was one of the reasons listed that Tumblr didn’t like Mising-e. This tool gives people the ability to chew up a lot more of Tumblr’s resources, but I’d be willing to enforce limits on Tumble Tree to get this within reasonable amounts if it means that it can be used at all.
I don’t mean to demonise Tumblr here - if I owned a blogging service of this size and traffic, I would disallow page scraping and excessive API usage as much as needed to provide a stable site too. However, I do hope that they consider things like this on a case-by-case basis, and perhaps work towards having this kind of functionality built in natively or allowed by 3rd party tools, as the response from launching this alone has proved just how much the Tumblr public would like a visual reblog mapper and notes filtering like this. If the difficulty for Tumblr is server load (which is likely), then this would obviously push any server-intensive features down Tumblr’s list of developement priorities considerably.
I hope this tree’ll pop back online soon, even if it has to have a severely reduced note limit - so stay tuned and I’ll let people know how this pans out. Thanks for your interest and support.