Comments on: Layers 3 and 4 http://mith.umd.edu/tile/2010/02/09/layers-3-and-4/?utm_source=rss&utm_medium=rss&utm_campaign=layers-3-and-4 Text-Image Linking Environment Fri, 23 Sep 2011 19:35:41 +0000 hourly 1 By: Ben Brumfield http://mith.umd.edu/tile/2010/02/09/layers-3-and-4/comment-page-1/#comment-11 Ben Brumfield Tue, 23 Mar 2010 17:00:05 +0000 http://mith.info/tile/?p=86#comment-11 The example I gave was pretty extreme, but I chose it because I wanted to point out the dangers in discarding user-created data as useless, irrelevant, or bulky. I've been <a href="http://manuscripttranscription.blogspot.com/2008/03/rails-logging-user-activity-for.html" rel="nofollow">tracking each user mouse-click in a DB table in FromThePage</a> rather than in user logs, and after a year of light use 1) there are fewer than sixty thousand non-spider records, and 2) I really wish I'd also captured the referrer field. My real concern is that imposing such a limitation in early-stage guidelines feels a lot like premature optimization. A lot of software projects (my own included) suffer from the <a href="http://en.wikipedia.org/wiki/You_ain%27t_gonna_need_it" rel="nofollow">temptation</a> to spend undue time and effort on scalability when they don't even have any users yet. In addition to the opportunity cost associated with this, I also worry that you're erecting barriers to adoption by encouraging sites to take on yet more work: constructing sandboxes and pruning/selecting user-generated data are not effortless tasks, and may be unnecessary. The example I gave was pretty extreme, but I chose it because I wanted to point out the dangers in discarding user-created data as useless, irrelevant, or bulky. I’ve been tracking each user mouse-click in a DB table in FromThePage rather than in user logs, and after a year of light use 1) there are fewer than sixty thousand non-spider records, and 2) I really wish I’d also captured the referrer field.

My real concern is that imposing such a limitation in early-stage guidelines feels a lot like premature optimization. A lot of software projects (my own included) suffer from the temptation to spend undue time and effort on scalability when they don’t even have any users yet. In addition to the opportunity cost associated with this, I also worry that you’re erecting barriers to adoption by encouraging sites to take on yet more work: constructing sandboxes and pruning/selecting user-generated data are not effortless tasks, and may be unnecessary.

]]> By: Doug http://mith.umd.edu/tile/2010/02/09/layers-3-and-4/comment-page-1/#comment-10 Doug Tue, 23 Mar 2010 15:08:36 +0000 http://mith.info/tile/?p=86#comment-10 Ben, I think that sort of storage would be great, and maybe possible, but if you think about how much space server logs can use (after four months online I think ours is already over 8GB) I think an individual server would be quickly overwhelmed if it tried to record every single interaction for a large set of high traffic sites. But your point is taken--at the moment relatively few users contribute content to scholarly sites (maybe more a problem of interface than interest, though). Doug Ben,

I think that sort of storage would be great, and maybe possible, but if you think about how much space server logs can use (after four months online I think ours is already over 8GB) I think an individual server would be quickly overwhelmed if it tried to record every single interaction for a large set of high traffic sites. But your point is taken–at the moment relatively few users contribute content to scholarly sites (maybe more a problem of interface than interest, though).

Doug

]]> By: Ben Brumfield http://mith.umd.edu/tile/2010/02/09/layers-3-and-4/comment-page-1/#comment-7 Ben Brumfield Tue, 16 Mar 2010 16:02:48 +0000 http://mith.info/tile/?p=86#comment-7 <i>Of course, storing the data of every user who ever used any of the collections in the framework is impossible.</i> I was really surprised and puzzled by this statement. Outside of the problems of spam, how much load from user interactions do you really envision? Even the smallest nugget of user-generated data can be useful, for example recording user click history can connect scholars who are researching similar things. Of course, storing the data of every user who ever used any of the collections in the framework is impossible.

I was really surprised and puzzled by this statement. Outside of the problems of spam, how much load from user interactions do you really envision? Even the smallest nugget of user-generated data can be useful, for example recording user click history can connect scholars who are researching similar things.

]]> By: Tweets that mention Layers 3 and 4 -- Topsy.com http://mith.umd.edu/tile/2010/02/09/layers-3-and-4/comment-page-1/#comment-6 Tweets that mention Layers 3 and 4 -- Topsy.com Tue, 09 Feb 2010 17:57:46 +0000 http://mith.info/tile/?p=86#comment-6 [...] This post was mentioned on Twitter by Editing Modernism, dougreside. dougreside said: Final two layers of 4-layer model for editions described on #TILEproject blog http://bit.ly/cNyiKj [...] [...] This post was mentioned on Twitter by Editing Modernism, dougreside. dougreside said: Final two layers of 4-layer model for editions described on #TILEproject blog http://bit.ly/cNyiKj [...]

]]>