Currently, the tags portal has 3 columns, and seems to be arranged as if Iāll read the first cluster in the left column, then the first cluster in the middle, then the first cluster in the right, then the second cluster in the left, etc. Basically, left to right then top to bottom (like in a book), rather than top to bottom in a column then on to the next column (like in a newspaper).
I say it seems to be arranged this way because thatās how itās alphabetically ordered and because thatās what would make āGlobal Catastrophic Risk (other)ā come after āGlobal Catastrophic Risk (artificial intelligence)ā (which I assume itās meant to, otherwise what āotherā means would be unclear).
But since each cluster is read top to bottom, I found myself naturally reading down to the next cluster within the same column, and so on, like in a newspaper. This meant I didnāt notice that the clusters were arranged alphabetically and instead just saw them as random, and that I was confused about why āGlobal Catastrophic Risk (other)ā seemed to come before āGlobal Catastrophic Risk (artificial intelligence)ā.
I think Iād recommend assuming the reader will scan the portal like theyād scan a newspaper, though Iām not sure.
The clusters arenāt in alphabetical order ā only the articles within clusters.
The clusters are arranged according to a couple of heuristics that I value about equally:
Try to make the columns of roughly equal length
Have the āotherā cluster near the bottom-right of the section (seems natural for that to be the last thing people look at)
Have related clusters close together (e.g. āeffective givingā and ācareer choiceā)
Iād prefer to have all the cluster names aligned horizontally, as on the LW Concepts page, but our extremely varied column lengths discourage that for now (this might change as we continue to add new articles, look at new ways to sort the page, etc.)
If anyone has an idea for making the page better-sorted and/āor more evenly arranged, Iām all ears. Graphic design isnāt my forte and the current version is quite rough.
The clusters arenāt in alphabetical order ā only the articles within clusters.
Oh man, youāre right, and I canāt see how I thought it was alphabetical except maybe that I noticed one case where one cluster came before the next one alphabetically and then didnāt check my theory. Gonna pin that obvious error of mine on jet lag after a long flight.
Those heuristics sound good to me.
As for aligning cluster names horizontally, it seems like this is already done for āOrganizationsā by just having (sometimes large-ish) gaps at the bottom of some clusters, and it seems like that looks ok to me? So maybe just do that?
And/āor you could do a vertical version of the ājustifiedā setting that Word and GDocs offers? I.e., you could have the vertical space between bullet points be larger or smaller depending on whether the cluster is less populated or more populated than those itās horizontally next to?
Currently, the tags portal has 3 columns, and seems to be arranged as if Iāll read the first cluster in the left column, then the first cluster in the middle, then the first cluster in the right, then the second cluster in the left, etc. Basically, left to right then top to bottom (like in a book), rather than top to bottom in a column then on to the next column (like in a newspaper).
I say it seems to be arranged this way because thatās how itās alphabetically ordered and because thatās what would make āGlobal Catastrophic Risk (other)ā come after āGlobal Catastrophic Risk (artificial intelligence)ā (which I assume itās meant to, otherwise what āotherā means would be unclear).
But since each cluster is read top to bottom, I found myself naturally reading down to the next cluster within the same column, and so on, like in a newspaper. This meant I didnāt notice that the clusters were arranged alphabetically and instead just saw them as random, and that I was confused about why āGlobal Catastrophic Risk (other)ā seemed to come before āGlobal Catastrophic Risk (artificial intelligence)ā.
I think Iād recommend assuming the reader will scan the portal like theyād scan a newspaper, though Iām not sure.
The clusters arenāt in alphabetical order ā only the articles within clusters.
The clusters are arranged according to a couple of heuristics that I value about equally:
Try to make the columns of roughly equal length
Have the āotherā cluster near the bottom-right of the section (seems natural for that to be the last thing people look at)
Have related clusters close together (e.g. āeffective givingā and ācareer choiceā)
Iād prefer to have all the cluster names aligned horizontally, as on the LW Concepts page, but our extremely varied column lengths discourage that for now (this might change as we continue to add new articles, look at new ways to sort the page, etc.)
If anyone has an idea for making the page better-sorted and/āor more evenly arranged, Iām all ears. Graphic design isnāt my forte and the current version is quite rough.
Oh man, youāre right, and I canāt see how I thought it was alphabetical except maybe that I noticed one case where one cluster came before the next one alphabetically and then didnāt check my theory. Gonna pin that obvious error of mine on jet lag after a long flight.
Those heuristics sound good to me.
As for aligning cluster names horizontally, it seems like this is already done for āOrganizationsā by just having (sometimes large-ish) gaps at the bottom of some clusters, and it seems like that looks ok to me? So maybe just do that?
And/āor you could do a vertical version of the ājustifiedā setting that Word and GDocs offers? I.e., you could have the vertical space between bullet points be larger or smaller depending on whether the cluster is less populated or more populated than those itās horizontally next to?