Meeting notes from OLPC HQ @Twine, 1400-1545
Discussions with other groups
Matt and Maryanne met w/ Megan Smith @ Google. They are most interested in helping, and pointing us to suitable contacts. Had a good meeting today with Google engineers here in Cambridge. Including Laura Cuozzo, Bernie Innocenti, David C?
Cynthia will spend 1.5h with Gates Ed. people soon. Matt has slides to share from last week.
They are currently looking for larger projects, each impacting at least 14,000 girls in at most 2 of 9 target countries (Afghanistan, DR Congo, Ethiopia, Kenya, Mozambique, Sierra Leone, Somalia, Tanzania, Zimbabwe). Projects will be funded for up to £30M for initiatives to be completed by 2016. (smaller pilot initiatives will be considered in a separate funding round in September) Concept note application needed by June 8.
Ideas for connectivity: you want to ensure there is wireless; may have to provide minimal infrastructure for it. That's nontrivial. But we can choose places that are more amenable than wonchi.
There was consensus that this seems like a good thig to pursue; the first round seems tailored to our work. Stephanie will follow up with Matt, who has long experience in this area.
Using the XO-1.75 touch?
- "OLPC XO 1.75" -- latest version of long line of laptops with same outside appearance. Internally, it is the same hardware used for the XO-3 tablet prototypes. Currently in production.
- "1.75 Touch" -- same laptop appearance as the XO 1.75, but with a touch screen.
- Will use a slightly more advanced processor (that's the current plan, at least)
- An Android port is planned; "a simple matter of software"
- In production by the end of 2012, customers already lined up.
- Advantage of being a known bit of hardware with kinks worked out, very scalable to high volumes, and robust. Free of some of the problems we've seen w/ the xoom.
- "XO-3" -- tablet form factor, same electronics as the XO-1.75
- No current customers.
- Might table a while to get the form factor as robust as the laptop form factor.
Plans for a series of apps
cscott runs down a list on the whiteboard
1. matching 2. abstract matching 3. letters - what should the last one be? m! a) sound -> letter, do we teach one or both? (do we say "ay. a, a, a, a. ay"?) b) letter -> sound c) sound/letter -> face 4. combinations a) cat/mat/at/cab b) tinkrbooks words/picture c) colors
(This was a recap of Literacy Project/2012-04-27_Notes#Contents_of_"Unit_1".)
running tests, using data
- we have a ton of data on tinkrbook, on what words people have seen a lot. we could throw the top 5 exposures to see if they recognize the word and a graphic, or just the spoken word.
- do we want to use the colors and numbers we are teaching in tinkrbook?
- better to view a couple separate tests, each standing alone.
- have one round of just words that they've seen.
- we could evaluate that every 5th time it runs, without trying to train off of it.
- abstract matching: could just be the word 'blue' and the colors. or a duck, with picture and word.
- if we are releasing one major new thing every week, it would be good to be guided by data. right now it takes 1 week to propagate a change, 3 wks later we get data on usage, if the change is noticed.
- we will see if anyone clicks through just because there is something new on the homescreen.
- next week we can try a more active alert about the new thing. if we discover (in 3 wks) they found it without that, great.
- Anansi is a spider figure they can relate to. you can draw it with a smudge or thumbprint.
- cf [ Ed Emberley]'s thumbprint drawing books
- you could add a leg each time you get a color right... have it form letter shapes... pulse when you get a letter right.
Future meetings and feedback
- An extended group retreat?
Maryanne - when can we just meet all day?
- Idea: a small retreat to figure out the big narrative, and who the characters are in the stories. There seem to be lots of great pieces but we should do this in a connected, unified way. We agree on the theory, moving beyond that to a [vision that the children will share]
- Meeting schedule
- Next Tuesday's meeting (May 15) will actually be from 2pm to 4pm+ at OLPC, and is for those who are working on a continuation of the linguistic-related issues stemming from the meeting before (see Scott's record).
- Then the following Tuesday (May 22), when Richard comes back, neither Stephanie nor Denise can be there, but the rest of us will meet and discuss Richard's findings. I would like all of us to hold that Tuesday time slot (1pm to 4pm+) for a potential hour or so with the Google group that Megan Smith is helping to organize. I don't know this moment whether such a plan will work, of course. But if they can also meet that day, then I would like us to meet first (from 1pm to 2:30pm) and then have them join us.... or vice versa at their offices (over Legal Sea Foods).
- Pre-reading for the meeting: all of Denise's memos, and Scott's minutes.
- On the overall narrative
- mw - 'I will never forget fardad's gorgeous village, and how it leads the way in...'
- scott - I want to schedule us too. We want that great narrative, but can set the hooks for it. Right now we can have a village faintly in the background of Chris's app. Then if we have time to develop [Fardad's village] further, they will recognize it.
- ff - Scott should show his hexagon (map). <shows a pretty demo, with hexagonal tiles continually flipping themselves.> Imagine the village as sharing a puzzle game. Each user is responsible for 5 pieces... imagine being able to flip a new hexagon when you complete something.
- Future ideas, testing them in coming weeks
- Can we converge our desire for a beautiful plan for 'tablet 2' onto something we can implement now for tablet 1, to get engagement data?
- scott - we should test more things out, making more mistakes.
- mw - let's get the top good ideas out first, so we can learn from that and put it into grantwriting. Do the standard things everyone expects we want to do. given time pressure, be sure to have our 4 best things up now; then you can try the other things.
- Let's say we are just handling simple text & matching, to know how it works, and get that deployed. This should include how we determine they are done with one thing and show the next set. Then we can move on (say to sounding letters). Talk to mw about this before moving on to that.
- For next week: can we get David Sengeh to record things for us? We will look into it, and into getting a variety of accents.