BumpTop

» see bumptop.com as it lived, on the wayback machine

~ the last words ~

the first time i saw bumptop was a youtube demo a hardware forum had linked, somewhere around 2008. it was a research video from the university of toronto, a guy on stage flicking documents around a 3d desktop like paper on a table. anand agarawala was the name in the credits, the project started as his masters thesis. the clip went around frmtr and donanimhaber with the kind of "this is the future of windows" framing the early ted talks always got.

this part i mostly watched from outside. anyway, the public release came on april 8 '09 as a download for windows xp and later, with a mac os x snow leopard build added later that year. the free edition gave you the physics and the basic stacking. the pro version unlocked the wall display and the photo grid widgets for about thirty dollars. the early reviews on lifehacker and engadget were curious rather than convinced.

i was not actually using it on my own desktop. for some reason, the install gave you a 3d room rendered in opengl with a floor and walls, your old icons turned into rigid bodies you could throw at the corners. you could pin a photo to the wall or pile receipts into a stack with a lasso. it was beautiful in a screenshot and irritating after about ten minutes if you actually needed to find a file (close enough).

on april 30 2010 agarawala posted that bumptop was being taken in "an exciting new direction" and would no longer be sold. two days later, on may 2 '10, google confirmed the acquisition. the price was not disclosed. agarawala joined the android team and worked on early home-screen prototypes. the bumptop product was end-of-lifed, the website served only a download link for existing customers for a while. in august 2012 google released the bumptop source code on github under apache 2.0 with no further plans.

~ the record ~

~ what it actually did ~

the engine simulated a small room with a floor and three walls (roughly). honestly, your icons became rectangular cards with mass. you could grab one with the mouse and throw it across the room. it would bounce off another stack of receipts on the way. the lasso let you draw a circle around several items, snapping them into a pile or a grid. the wall accepted pinned photos. there was a sticky-note widget you could drop on the floor.

the part that actually had a use case was the auto-pile feature. drag a set of files onto a folder icon and it spread them in a fan to preview the lot before opening one. mostly, the gpu requirements were modest on '09 hardware but the cpu got loud when the room had a few hundred icons in it.

~ the google deal ~

this part i was not around for, mostly. either way, google bought bumptop one week before google i/o 2010. the wider press read the deal as a tablet move, the assumption was that android tablets needed a more spatial home screen than the icon grid that shipped on phones. about eight months after the acquisition google posted a sneak preview of android 3.0 honeycomb showing a 3d home screen, and the press connected the dots back to bumptop.

honeycomb was not bumptop, the 3d effects on the tablet ui were cosmetic rather than physics-driven (or roughly). agarawala worked on the android home-screen team for a few years. the bumptop brand never reappeared as a google product. the company folded into the larger android org with a quiet end-of-life note for paying customers.

~ what got buried with it ~

the 3d desktop as a category was already losing in '10. anyway, microsoft had shipped looking glass and project cool versions of 3d window switchers earlier in the decade with low adoption. compiz had wobbly windows on linux. apple's expose was the version most users settled on, a flat zoom-out grid with no physics. bumptop's bet was that physics would land if you made the room small and the icons big. it landed in a demo loop and not on people's actual desktops.

the source code drop in august 2012 was the kind of acquisition coda where the acquirer dumps the source as a goodwill gesture and never touches it again (close enough). the bumptop repo on github is still there in 2026 with no commits since the initial push. anand agarawala stayed at google through the android years before leaving in 2018 for spatial.io, a vr workspace startup that pivots back toward 3d productivity ideas. the thesis project did most of its cultural work in the demo loop, the shipping product was the part that did not survive.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »