Electric Airplanes

Imagine: an electric airplane that could be recharged by lightning bolts!

Chrome Webstore and Entrypoints

I've been randomly thinking about why Chrome has N hundred million users, but some of the popular Chrome Webstore apps like Bejeweled only have 1 million users -- at least a hundred fold difference.  Maybe people really only think of Chrome as a web browser, and don't bother with the app store.  Without people using the app store, and talking about it, there aren't as many entrypoints, and thus fewer chances to pull in more users to repeat the cycle -- a pretty obvious point.  I don't offer any solutions in this post... just posting some thoughts.

To me, the Chrome Webstore has one entrypoint: the new tab page on Chrome.

I've seen ads for the chrome webstore, so I guess that is a different entrypoint, except that it isn't all that different.  It leads to the same spot, the landing page.

These entrypoint links don't highlight any apps in particular -- they rely on the "Featured" section of the webstore landing page.  If none of the Featured apps catch your eye, you'd probably leave.  If you were patient, maybe you would give it another click and dive into one of the categories, like games.  I don't know how many people would really just browse the webstore, because that takes too much time and the list of apps don't change often enough to warrant browsing the store.

How does this compare to other app/online stores?  I've never actually used the iPhone app store, but I've used Android Market, Steam, and of course Amazon.  Between all these stores, really the key is that there are lots of people "linking" to specific apps, and this gets people going to the store. You know what you want. There's no effort.

Sometimes these links are ads, but hey, as long as you know what you think you want, you'll go get it.

Amazon though, is so big that it is a bit special in two ways. (1) I would actually go to Amazon, just to use their search (e.g., to look for toys, or even to look for reviews for products). (2) Amazon gets tons of links from Google search and from product review sites, etc.  I.e., an extended version of word-of-mouth.  The CWS just isn't big enough for any reasonable user to think of doing (1).   For (2), I'm actually not sure that the CWS gets any hits at all from Google search, but maybe the issue is again that it's not big enough and doesn't have enough links to the CWS to make it relevant. (Edit: it looks like it can show up in Google search results: Bastion's CWS entry shows up on the second page of search results).  For example, Mini Ninjas doesn't get any hits for the CWS.  Wikipedia, Amazon, Steam, and Gamestop are on the first page.  The Mac app store shows up around the third.  The rest are FAQs and reviews.  That ranking seems to make sense based on popularity.

Education => Technology, Technology => Education

"Technology is the art of doing well enough more cheaply." - David Packard

I came across this quote while reading through a lively discussion about Khan Academy and how online educational resources should not replace universities, but supplement them. The beginning of this post borrows from that discussion. I bring up some ideas in the second half of the post, but I'm not an educator, so my idea of areas for improvement might be way off.


I think the Dave Packard quote works well with some technological "advances", e.g., factory farming. Though, I don't think it is the intention of Khan Academy, MIT Open Courseware, or any of the other online university-ware, to turn our schools into factory schools. They are intended more as supplements.


Going back to the quote, one important bit is, who determines what is "well enough" for things like education, which is a long term investment, while it seems most policymakers simply want to look good in the short term? There is also the issue of "more cheaply". Educators are already cheap (not paid as well as most professions), but there are still not enough.


What about kids who would rather watch YouTube videos of kittens, than educational videos? I.e., does this really help a broad audience, or does it simply help the already-curious and bright children learn more? One father/professor noted that even bright children would rather play xbox than read about your-favorite-theory. The same professor also noted that curiosity and motivation to learn often arrive slowly over time. Perhaps additional sources of educational-encouragement (e.g., Khan Academy) will help stir that up, but only if it can compete for children's time against other technology (e.g., video games, Facebook). 


Actually, perhaps tech-enhanced-education does not need to compete over the same share of time! Tech can supplement and improve classroom time, and leisure time can remain leisure time, and if classrooms do their job then, over time, learning becomes leisure.


For learning as leisure, there are already resources like fun podcasts, museums, puzzle games, etc., so let's focus on the classroom bit.


What are some of the inefficiencies of classrooms? I've seen University experiments with tools for helping super-large lecture halls (200 students!) conduct question-and-answers. In this post, I'm not interested in those settings but am instead interested in grade school settings with "small" classrooms.


There is currently no way for one teacher to talk one-on-one to every child in the class for long. Could tech help identify children who are having similar issues and allow a teacher to address them more efficiently?


It currently takes a lot of time for a teacher to really get to know each child and really know how to teach them. Could tech observe children over a long period of time (years!) and help identify habits of children and identify strategies for reaching out to them? It is usually only the parent that has this privilege, as teachers change every year. This could provide a dossier of each child as they move from one grade to the next.


What are other inefficiencies of the way current classrooms work?


Of course, these ideas are all flawed, but we can discuss these flaws and discuss some more:

• Most of these skills of observing and understanding people are very human, and out of the reach of computers. 
• The thought of observing a child, and possibly providing teachers with biases from dossiers is quite scary.
• These ideas all talk about "observing" and "identifying" characteristics of children. How? Universities are already interested in identifying characteristics of children (or at least "aptitude"), and tests, as they are currently designed, are already controversial.
• It might "take 10 years to really know a person". By 16 (assuming observations start at age 6), a child will almost be an adult!

What else is possible:

• For the issue of long-term observation: Have friends, or parents observe. They are trusted (sort of?), and are already interested in knowing the kids. This is sort of like the peer feedback system at Google (but that's for big kids, and your peers aren't really your friends). Alternatively, perhaps have teachers teach the same group of kids for more than a year. I remember taking two consecutive years of math with the same teacher. The first semester was scary and the second year was fun and exceptionally clear (though the class was also much smaller that second year, and much of the material was review...)!
• For the issue of clustering and more efficiently addressing similar issues: perhaps technology can help provide a broader set of "regular" assignments (though cheating would be an issue) and could make grading automatic. Grading automatically will at least help teachers skip the boring part of assignments and tests, and focus on where misunderstandings occur and how to address those misunderstandings. The automated grading can then also automatically provide ranks of the toughest problems and frequencies in which people have trouble. These ranks can be shared across schools and teachers and share tips among each other over a social network or (umm) mailing lists. For harder-to-automate assignments like essays, software can do the boring parts of grading, like grammar and spelling (well... we all know how good Word's grammar checker is...). I'm not sure what else natural language processing and machine learning can do with essays.

Well, "hai kow lah" (that's it) for now.