The UX (both keyboard and software) was .. just awful.
The software + hardware was really poor, even for the time. (Even the original Raspberry Pi is significantly faster -- I know this isn't a fair comparison, but just for reference)
Also a lot of the OLPC functionality didn't work reliably, or didn't work at all. I bought it specifically for the mesh networking, which was cut entirely. The devices did work over WiFi, but would struggle to see other OLPCs over wifi reliably, even when connected to the same AP. I spent a lot of time reading forums online to troubleshoot and download fixed drivers and such. I remember explicitly wondering "how are children in poor areas who depend on mesh networked internet, supposed to be figuring all of this out?".
The screen was hard to use. The e-Paper mode was nice, but the regular mode was blurry/grainy and slow and dim and just difficult to see. In terms of Software + UI + responsiveness, it felt more like a really big Palm Pilot, and less like a laptop.
---
I loved the mission, I loved the ideas, I loved the design. The exterior of the case really was durable and kid-friendly. I loved the idea of the screen. I threw them more money than was reasonable. I was just fairly underwhelmed by the device as it actually shipped. It felt like a prototype of a dev kit, which is fine. But it was sold as something for children to use, and at least at the time I messed with them, it was nowhere near ready for that.
• The hardware was not designed in conjunction with the software; it seems to have been built based on a "wishlist" of features drawn up before the software was developed. There were a number of major hardware features which were unused or unsupported by software, including but not limited to the resistive tablet, the low-power-mode ("e-reader mode") for the display, and even several of the hardware buttons (including most of the ones next to the screen, which were mapped confusingly to arrow keys).
• Aside from being confusing, the software was very rudimentary. The two major activities which would be useful to students (the web browser and text editor) were not actually Sugar applications at all, but standard Linux applications (Firefox and Libreoffice, iirc?) wrapped in their interface. There were very few nontrivial native Sugar activities available.
• The Sugar interface, as designed, had some very strange semantics which were not intuitive, and not reflected well in the user interface. Every activity opened by the user was saved in the filesystem -- even ones which were not useful to save, like web browsers or simple memory games. The OS would automatically discard old activities as necessary to keep storage available -- so recording a long video, for instance, would delete older documents created in the text editor.
• Development on the project had a lot of bizarre priorities. For a time, one major development priority was to build a reverse-engineering suite for the OLPC, so that students could reverse-engineer the mesh networking firmware and develop their own under a free license. Fortunately, I don't think this got to the point of actually being developed.
You're definitely right on that. The hardware took long enough to design, but the software wasn't even done when they started shipping.
> the resistive tablet
For people not familiar with the XO-1, it originally included a weird touchpad. The center third was capacitive, and would work with fingers. The Entire pad was resistive, and would work with a stylus (which was never shipped).
The dual resistive/capacitive was unreliable [0], and was eventually replaced [1] with just a capacitive touchpad with the same area as the original touchpad.
> the low-power-mode
This was a real shame to leave out. The OLPC XO-1 includes a specialized display controller [2] which can drive the display even while the CPU is suspended. THe idea is that the CPU would render one ebook page at a time and sleep between page turns. Unfortunately, the software for this was never implemented. All the rest of the hardware is there; it would have been really cool to have a near-zero-power ebook mode!
> several of the hardware buttons (including most of the ones next to the screen, which were mapped confusingly to arrow keys)
I'm not quite sure what you mean here. All of the keys on the keyboard are mapped [3]. The only ones I can think of which aren't used for much are the progressive dots (F5-F8), which are intended for application-specific use [4].
The buttons around the screen are mapped to the arrow keys (on the left side) and Home, End, PgUp, and PgDown (on the right side). I really like this, because when you fold the screen around into tablet mode, you can easily navigate a document. Reading a PDF or tall web page this way is a pleasant experience.
> the web browser and text editor were [...] standard Linux applications (Firefox and Libreoffice) [...] wrapped in their interface
You're right. The web browser is Firefox with a bunch of XUL stuff, and the document/text editor is AbiWord.
> The Sugar interface, as designed, had some very strange semantics
The idea behind an OS that journals everything you do with it [5] is great. All the power of a VCS, applied to everything you do on a computer. Unfortunately, the implementation in the OLPC is crude and severely hardware-restricted.
> development priority was to build a reverse-engineering suite
This is actually kind of funny, but the only mention I could find is in [6], which is a collective braindump page for software that would be cool to have on the OLPC. Do you have links to any discussion on this?
[0] https://wiki.sugarlabs.org/go/0.90/Notes#OLPC_XO-1_touchpad_...
[1] https://web.archive.org/web/20160825191424/http://lists.lapt...
[2] http://wiki.laptop.org/go/DCON
[3] http://wiki.laptop.org/go/Keyboard
[4] https://wiki.sugarlabs.org/go/Human_Interface_Guidelines/The...
For what it's worth, I was the OLPC employee working the most on the software for this, and it worked and shipped. I recall that there were serious hardware bugs that went unfixed until XO-1.5 and XO-1.75, but just those two models add up to millions of deployed laptops in the field that were using this CPU-off-screen-on mode when idle. If you've ever seen the power LED off or flashing on an XO while the screen is on, it was using this DCON mode.