In Jakob Nielsen's comments about the Internet userbase exceeding one billion, he says:

The Mac, for example, already matters less than you think. Although it has a prominent role in the U.S., it's hard to refer to a company with single-digit market share as "dominant." In Asia, the Mac is practically nonexistent.

I disagree with his assessment. Apple has a noticeable presence and market share in Japan, and at one point, something like 55% of Apple's international sales were from Japan. The only reason why Apple has practically no marketshare in the rest of Asia is because most of Asia (population-wise, that means China and India) can't afford Macs. But with economies modernizing in those countries and as disposable incomes increase, I think there's a very good chance Apple will see significant growth in those markets. [Tagged: macs asia jakob+nielsen ]

A week or so after I bloged about my dream search box, Yahoo! internally released a new search feature called Open Shortcuts that was functionally equivalent to my hack (the timing, AFAIK, was completely coincidental). Well, as of yesterday, this feature has apparently become publicly available. Read more over at the Yahoo! Search blog.

Despite working at Yahoo! (though not in the search division), I admittedly still use Google as my default search engine. For me, the main reasons that I haven't been able to switch (I've tried, I swear!) are:

• Google's result are marginally better (more relevant, less spam) than Yahoo
  
• I like the weather forecast display better on G
  
• I like the stock ticker display better on G
  
• Google's calculator is better (try: "0xef to binary","e^4", "14!" --none of these work on Yahoo!)
  

Open Shortcuts is kind of cool because it does lower the barrier for conversion. I can set my default search engine to Yahoo! and selectively send queries to Google with 3 extra key strokes. Or I can send my stock symbol lookups to Yahoo! Finance (which, at the end, is usually what comes up as the #1 result on Google anyway). But then, 3 extra key strokes to lookup a 4 letter ticker symbol adds 60% more work... and that's a lot.

The other cool thing about Open Shortcuts is that it adds stickiness to search. Since the shortcuts are completely customizable, once you've set things up the way you like them and get used to it, you're likely to keep using the product. Whereas with generic non-customized search, the cost of switching is practically zero (assuming equivalent functionality and quality). [Tagged: yahoo search open+shortcuts ]

Lots of talk about the death of IE5 for Mac (well, more like 2nd death, the 1st death was when MS halted development), and the general tone seems celebratory. Now I'm anti-MS, anti-IE, yadda yadda yadda as the next, but I think people are losing context here.

*dreamy transition sequence*

It's a long time ago. Year 2000 or there abouts. MacOS X PR1 comes out to revolutionize the platform, to bring the Mac OS out of the stone ages. Guess what browser it comes with? Yup, MSIE 5. And guess what? It worked. Safari wasn't out yet, neither was Firefox. Lots of people were still using Netscape 4.7, and Netscape 6 was one slow buggy hunk of software. Frankly, for a while, MSIE 5 was the best browser on OS X.

*dreamy transition sequence*

Sure, by today's standard, IE5 is a crappy browser no matter how you look at it. But at it's moment of death, let's at least acknowledge it for what it was. At the very least, the developers who fought an uphill battle to get it out deserve it.

I'm at BarCamp right now, kinda kicking myself for not having come out here earlier (I got here a little after 4pm today). So far I got part of a session about SVG, and a service called Pandora which is basically a music streaming service that makes recommendations based on music you like (and finds matches from less well known artists --the so called "long tail"). One interesting little tidbit I got out of the Pandora session was about naming; apparently nonsensical names are memorable than the sensical ones. Right now I'm catching the tail end of the Women in Tech session where, it sounds like they're talking about creating an Open Source mobile-based geolocation-yummy service thinggy. [Tagged: barcamp ]

Just saw some details about PS 3, and felt rather sick:

At the heart of the PlayStation 3 lie two very powerful processors. One, the main processor, is a new chip called Cell that was developed for use in the console by SCEI, its parent company Sony Corp., Toshiba Corp. and IBM Corp. The chip is based on IBM's PowerPC architecture and will have seven cores, each running at 3.2GHz, to give the chip a total performance of 218GFlops.

Running alongside the Cell will be a graphics processor developed by Nvidia Corp. that is capable of handling full high-definition images and boasts performance of 1.8TFlops. Together, the two chips will give the PlayStation 3 a total system performance of 2TFlops

Maybe it's because of the way I was raised (no comics, no TV, certainly no game consoles) but unlike most people my age, I'm distinctly anti-gaming. At the very least, I tend to clump computer games in with sitting on your ass watching TV and other such mindless (mind damaging) activities. There's nothing wrong with it if that's what you want to do to yourself, but I find it an aweful waste.

So Sony is going to ship millions(?) of these high-performance computing units, which will spend an aggregate several hundred million hours of computing power, for uses that average out to (IMHO) negative human development. Imagine what it would be like if all these gamers went outside and played or otherwise did something useful/not-unhealthy, and used all that computing power on something useful. Imagine all the good that could do to mankind...

But, I think I'm just going to have to hope that I'm wrong on this one, and hope that there's some good in computer games. I mean, we wouldn't really spend so much time, energy, money, and brain power on something that's self destructive... would we?
[Tagged: ps3 ]

For as long as I can remember (i.e. at least a decade), I've been the go-to guy for Mac related issues for friends, family and people of that ilk. Recently, I've been telling people to just go to the Genius Bar in the Apple Store downtown. Skip the campus computer store (they're useless), don't call Apple (half the time they make you run in loops); just take it to the Apple Store.

The amazing thing is, most of the time, they come back and tell me good things about their experiences.

[Tagged: apple ]

I was talking to my friend Yitz the other day (whose assbook has taken on a post-Scav life of its own¹) about the possibilities of Open Source social networking software.

In some sense, Open Source social networking software is meaningless, because social networks derive value from their members more than from the underlying software. Slashdot and Livejournal are both built on open source software, but the sites and their respective communities (I would argue), have made greater impact than the software themselves.

However, as people begin to realize that just about any software created these days, particularly for the web, has to be social networking software of one form or another, an Open Source social networking library that provides core functionality could become valuable. Yitz created a site that does most of what The Facebook does (and some things that it doesn't do) in about a week, but most people don't have the time, energy or brains to do that. Even if they did, it's a royal waste of time to recreate 90% of the{face|ass}book just to do something a little different for the remaining 10%.

One application we discussed was the marriage between OpenHive, and a social networking core. At this point, the assbook has more extensive social networking capabilities than OpenHive, and it makes sense to take those core SN capabilities and simply add library sharing features on top of it.

Another interesting possibility would be if the OSSN library had externally accessible APIs, that allowed different social networks built using the same core to communicate with each other; effectively building a grid of social networks (or a network of networks). So, for example, if you have an account on The Assbook, you wouldn't have to re-register for OpenHive or rebuild your friends list, and vice versa. (side note: FOAF/RDF might be useful here)

Of course, if various services opened up a little more, there wouldn't be much of a need for an open source interoperable social network-network. For example, OpenHive accesses data from two external sites: The Facebook, and Amazon. We did nasty screen scraping from The Facebook. We used well documented APIs for Amazon. I'm sure you can guess which we preferred.

I didn't really plan it this way, but I think I'm saying what I said in my last post: OpenSource is great, but open APIs are better.

¹Some numbers on his blog, also see articles in our school paper and Stanford's. [Tagged: social+networking open+source open+APIs theassbook ]

I think Lawrence Lessig talked about something similar in Code, but we're having an interesting discussion in the MacLab.

The problem is this. Our machines rev nightly, and as part of the revving process, we've been wiping all user data. Our explicit policy is that we make absolutely no guarantees about user data, particularly over night. However, the reality of it is, most users are not entirely aware of our policy, and have a tendency to ignore posters, warnings, etc we put up. One of our tutors modified the revving process to not delete files that have been modified within 3 (or any fixnum) days.

I think this is a horrible idea, but others seem to disagree.

The reason I think this is a horrible idea is simple: we are creating conflicting policies. One policy is our explicit one, where we say we don't save user data over night. The other policy is the implicit one, where we in practice will be saving user data for some number of days. Where this becomes an issue is that users are more likely to follow our implicit policy than our explicit policy, because even though they do not read signs, they definitely notice when they see (or don't see) their files from previous days.

So we will then, through our own practice, be implicitly telling our users that user data is stored for some indeterminate number of days, while not being able to provide the level of service/security users will naturally expect/assume (that is, user files can suddenly disappear, and we will simply point at our unnoticed signs).

What really baffles me is that, none of our users are asking us to store data. If they ask and we tell them we wipe the hard drives every night, they understand. Yet by introducing an implicit, unofficial, and un-demanded service, we're jeapordizing users data. It makes no sense to me.

What am I missing?

This year I've been working mostly with "manly con"¹ so I've been outside most of the time, and away from computers, email, internet, etc. I took on the trebuchet item, and lead the design and construction of a kick-ass trebuchet with a 10lb counter-weight (the item specifically states that the counter-weight must be <=10lb; otherwise we would've made it bigger). We think we're supposed to fling calculators at a model building with it, so we're going for accuracy over range... I think we'll do well on this one.

Anyway, I've always wanted to make a trebuchet, and now I have.

¹:Stands for manly construction, which does big construction projects with power tools, as opposed to "girly con" which is in charge of artsy craftsy things and making things we make look nice --membership to either cons, however, isn't gener-specific in any way). [Tagged: scavhunt university+of+chicago trebuchet ]

ScavHunt is on. You'll hear about it on Slashdot soon, but one of the items is to slashdot slashdot. Any ideas? Also, we need to change Google's logo on judgement day. Again, any ideas? [Tagged: scavhunt university+of+chicago ]

I'm sure this is a common problem everybody has, but say we want to do the following:

1. Have a different password for every service
2. Not have to memorize all the passwords
3. Not have to write down or otherwise store passwords
4. Have passwords readily retreivable (without asking the service)

... and have passwords be secure even if an attacker

1. knows the method used to generate passwords
2. may have key loggers installed on any computer you can use
3. knows everything about you that's publicly accessible
4. knows some, or even all, of the passwords except the one he/she wants

So this is more of a fun puzzle than a real-world problem, although such a system could obviously be useful for the paranoid. Anyway, some friends and I were thinking about this, and we got distracted before getting far, but came up with some thoughts:

• We obviously need to use some form of secret information that's unknown/unknowable to the attacker
• One idea is to come up with a one-way function that's fairly simple to compute given your secret information, and apply it to another known string (your login name, or service name) on the back of a napkin (so we're assuming the attacker isn't around -- and you'll have to eat or burn the napkin)
• The problem with the previous algorithmic solution is that, whatever you can do on a napkin, a computer can do millions/billions of times faster, so it's potentially suseptable to brute force attacks (unless you can do arithmitic with 100 digit numbers on the back of a napkin)
• So another idea is to use information that humans (or more specifically, you) can easily store and retrieve, but would be very difficult for computers to figure out. For example, childhood memories, word associations, emotional things, graphical memory, etc.
• One possibility might be to memorize one word or idea, that leads to additional data. This might be the name of a song (to come up with the lyrics), chapter+verse of the Bible (if you happen to be one of those people who can quote the Bible verbatim), etc. The only requirement is that you have to be able to recite the data without error, and it has to contain most (preferrably all) letters of the alphabet. This can act sort of like a one-time pad.
• From there, you can do a number of things, like write down the text in a grid, and use grid coordinates to go from say, the service name to a sequence of numbers (which you may need to map back to a larger aphabet).

Any other thoughts? [Tagged: passwords cryptography ]

Finding a good name is probably the most difficult thing about starting a new project, but also quite vital (word-of-mouth doesn't work when someone says "I found this really cool thing... but I can't remember what it was called!").

Anyway, over the years, I've made some observations on good domain names, and I present the following conjecture.

Ryo's Conjecture on Good Domain Names:

• A domain name can be sensical or non-sensical.
  
• A sensical domain name must be:
  
  
  
  • comprised of no more than 2 common words (or word fragments)
    
  • which are memorable and easy to spell
    
  • at least one of which is relevant to the project/service/site
    
  
  
• A nonsensical name must be:
  
  
  
  • Preferrably less than 6, no more than 8 letters long
    
  • Be spelled pretty much exactly how it sounds
    
  
  
• The Dotcom might've been a bust, but .com is here to stay; if you can't get the .com domain, forget about it.
  

I heard about Rojo for the first time just last night, and it appears they just launched their public beta today. I just created an account and I've only spent a few minutes tinkering with it, but I think it's got the best UI of all the web-based aggregators I've seen. The design is clear and simple, but functional and fairly intuitive.

Only major complaint I have is that, for a service being rolled out today, it uses surprisingly little (i.e. none, as far as I can tell) Ajax/DHTML (for example, there's no excuse to reload a page simply to add a tag to a feed).

Speaking of web-based aggregators, a friend of mine wrote a prototype XSLT front-end to IlohaMail's feed aggregator, which takes advantage of the fact that IlohaMail can re-publish aggregated feeds as a single Atom feed. [Tagged: rojo feed+aggregators ]

A guy walked into the MacLab today:

Guy: I'm trying to get some files off my old Mac... do you have any suggestions?
Me: Does it have an internet connection?
Guy: No...
Me: Which model do you have?
Guy: I'm not sure... but I bought it in 1992.

That's gotta be the oldest personal computer I've ever heard anyone using. I mean, 1992 means it's probably one of the Mac II series or perhaps a Quadra. Hell, PowerPC didn't come out until 1994. He must've paid a premium for the machine then, but he sure got his bang for his bucks. [Tagged: Mac ]

GoogleMaps rolled out (as expected) a satellite imaging feature. Holly shit, it's soo coool. (I am here).
[Tagged: Google+maps ]