I'm still not sure who I'm going to vote for in the upcoming New York mayoral election. On one hand, Republican candidate Michael Bloomberg is not bad, and in fact has done some great things for New York, notably the smoking ban and 311. On the other hand, he's a Republican, has publicly supported Bush, and seems to be ignoring the subways. Fernando Ferrer, the Democrat, seems quite good too, though. On the other hand, he's an unknown quantity.
Then there's Jimmy McMillan, who, according to my voter information packet, is from the "Rent is Too Damn High Party". I quote from the packet:
What is the most important issue in the city you would address if elected?RENT is Too Damn High there is nothing else to talk about. All poor people are being ran out of New York.
What other important issues would you address if elected?RENT Is Too Damn High there is nothing else to talk about.
The guy's got a point.
Well, so, I may vote for Republican, which I haven't done in a while. Strangely enough, the other Republican candidate for Mayor, Thomas V. Ognibene seems quite reasonable as well, even complaining about the "laissez-faire" attitude that spawns big-box retail in the area. I guess in New York, Republicans are just another kind of Democrat.
For the mayoral election, it really doesn't matter who I vote for anyway. Bloomberg will definitely will big.
Another weird thing: Bernie Geotz is running for Public Advocate. You can check out his website, which has a list of his plans to encourage napping, promote vegetarianism, and other ideas, some more well thought out than others.
Wikipedia has problems. Big problems, even acknowledged by it's founder, Jimmy Wales. The problem is that Wikipedia styles itself as an encylopedia, but it's articles are of questionable accuracy, and are often badly put together.
The strength of Wikipedia is that anyone can add content. This is a nice way to get a lot of different content, with a lot of different perspectives. However, it is not a nice way to get a coherent and valuable article.
Wikipedia needs to have, first of all, a reputation system. Wikipedia makes no effort to distinguish between valuable contributers, and contributers who do nothing really worthwhile. Sure, vandals might be banned, but everyone else is equal. Also, reputation has to mirror real-life reputation. If you are writing an article about the Marvel Comics, the input of Stan Lee is just more valuable then the input of your average Wikipedia user. But note that in this case the reputation is constrained to a particular topic. I trust Stan Lee to help create an article about Marvel Comics, but not to help create an article about differential equations. Similarly, some posters might create value in a particular topic, but have a negative effect in another topic.
So, the first element in my proposal is a reputation system where each user has a reputation in a particular topic area.
With this reputation system, you could have three levels of content for each article. The first is the content as it exists today, which is the level-playing field free-for-all. I call this content level the "low level content". Users of high enough reputation (which should be just a handful of users) get a brand new article in which they can build up, using new content, or content from the lower level of content. This content level is the "medium level content". Finally, what should be just two or three users with the highest reputation should edit a "high level content", drawing from the lower two levels.
Each level provides ready-made text which to edit to the level above, and since each successive level is contributed by fewer people, the editing should be tighter, and the site more stable.
Notice how this solution is different than the solution proposed by Wales, which is "branching" content that has reached a level of acceptability to a sort of "release branch Wikipedia". That solution results in a static release branch, which is not very internet-appopriate because it is static. Also, it assumes that content can get to a releasable quality with the current system. That is almost certainly a false assumption. My system lets Wikipedia be Wikipedia, but also provides for additional layers of quality that are less and less Wikipedia-like, and have more and more quality. All the levels can feed off each other, and I think the entire process has something for everyone.
The question I have not answered is how to do a reputation system, especially a reputation system that is context-sensitive. It's a difficult question, but I'll try and write another article about it soon. I have to figure it out first, though.
There was an interesting article in the New York Times a few days ago called Meet the Life Hackers, which described the study of productivity. It had to do with a lot of stuff I'm interested in. It had a bit about the human factors of productivity, a bit about software engineers, and a bit about interruptions.
The bit about productivity got me thinking. There's areas where I am pretty good, and some where I am not so good. First of all, I am rarely bothered by excess interruptions as work. Much of that is the somewhat isolated nature of my job. But also, I check mail in emacs, which is where I do most of my other work as well. And I never figured out how to get emacs to tell me about new mail. So I only get new mail when I remember to check for it. Sometimes I check every few minutes. Sometimes not for several hours. But I'm never interrupted.
Emacs also came to mind when reading the article, which tells how technology writer Danny O'Brien asked the 70 most prolific people he knew about how they managed their life. One of the things he found was that:
None of them used complex technology to manage their to-do lists: no Palm Pilots, no day-planner software. Instead, they all preferred to find one extremely simple application and shove their entire lives into it. Some of O'Brien's correspondents said they opened up a single document in a word-processing program and used it as an extra brain, dumping in everything they needed to remember - addresses, to-do lists, birthdays - and then just searched through that file when they needed a piece of information.
Wow, that sounds just like how people use emacs! Except that emacs isn't a "simple application" by any means. However, I've been keeping a running list of notes in emacs for years now, and it works extremely well. For instance, I store the SQL statements I need to do a common task at work. When I need it, I just open up a buffer, do an incremental search for the relevant words, copy, and paste. It works extremely well, and I store work-related items, places to try for lunch, and anything I want to be able to recall quickly, but don't want to actually remember.
One thing that hasn't been working so well for me is keeping a "todo" list. I started by having a text file I edit in emacs, with everything I needed to be working on, along with all notes related to those items. However, this was hard to keep up-to-date, and became cluttered with items that were stalled and had voluminous notes, making the important items hard to see.
Today I installed planner.el, a way to organize todo items into projects. It's very popular, some people even learning emacs just to use it! You can see items on a todo list for the current day, or see them grouped by project. Each day, or each project can have notes and schedules attached. It's all done in a emacs wiki system, which is publishable as HTML. So far, I'm very happy with it. I think this might be the final part of my emacs productivity puzzle.
I love a restaurant that knows what it does best, and sticks to it. Nick's Pizza, for example, in Forest Hills does pizza and very little else. Similarly, this week my co-workers and I went to a place that serves hummus, hummus with chickpeas, and hummus with fava beans. Oh, and a tomato stew. That's it.
The hummus was very good. It was so smooth, and they put lots of different spices in it so that it keeps your interest. And there's egg in it. Everything in this restaurant had the capability of having an egg on top. I wish more restaurants were like that.
Not only was the hummus very good, but the pita bread was excellent, and obviously freshly made. I really enjoyed this bread.
I did not have the hummus, however. I just tried someone else's. What I had was... well I can't remember the exact name, but it was a tomato stew with some vegetables. And a fried egg on top. Yes!
And if that delicious meal wasn't enough, there was Turkish coffee. I love Turkish coffee!
All in all, a very enjoyable experience. I'll be coming back here.
Mac OS X's user interface for most things is usually pretty intuitive. However, the Finder is something I have some problems with.
First, there's the whole spacial-mapping thing, where folders are laid out spacially, requiring you to actually move folders around so they don't land on top of each other. I know UI gurus are big fans of this, but I just don't get why. It seems like much more work for no apparent gain except in very specialized circumstances which I have yet to encounter.
Second, and more annoying, there's the occasional thing you want to do, that seems quite reasonable, but just baffles you. Like you are in a folder, and you pick up an item and want to drag it to your parent. Is there any way to do this without dropping it and switching the Finder view to either show parents, or opening up a new Finder window?
Another Finder problem bit me recently. I was trying to copy a file from my wife's account to my account. I thought I would be all Mac-like and use the Public Drop Box, a folder for each user that exists solely so that users can drop files there. However, my wife is not an admin, and I just could not get to the parent of her home directory. Therefore, there is no apparent way to navigate to my drop box. I also couldn't find a way in Finder to just jump to a folder. Eventually I cheated and opened up a Terminal and typed "open ~ahyatt/Public/Drop\ Box", but it felt wrong. I must be missing something here. I can't believe the Public Drop Box is not accessible.
Then again, the Mac is great for sharing things between computers, with "Bonjour" and various other technologies. But sharing things among accounts on computers? It's tough as hell, man! And it makes no sense that I can share things with another computer easier than I share things within the computer.
Whoops, it looks like this little rant has ended up on a completely different topic than the one I started with. Sorry, folks!
I've heard good things about Sweet-n-Tart, a nice Chinatown Cantonese spot with dim-sum type items. I tried it out with a friend last week, and thought it was pretty good.
We had some standard dim-sum type dumplings, along with a slightly unusual shrimp & asparagus dumpling. We also ordered chicken and clam congee, a strange combination that worked better than expected. Also, we had a beef noodle soup. The beef flavor was unexceptional, but we liked the noodles.
All in all, a nice meal, and it wasn't too expensive (although it was expensive by Chinatown standards). Nothing was greasy, and everything tasted fresh. I'd go back, especially since there aren't that many good Cantonese places around New York.
I previously blogged about Flushing restaurant Spicy & Tasty. I went back and got some different dishes this week. As usual, it was all very good. This place uses real Sichuan peppercorns, which means that not only is the food spicy, it also has this strange mouth-numbing taste. Pictures after the jump.
So, I got a certain type of boiled dumpling, soaked in red chili oil. I love this type of dumpling, and this one manages to be elegant in appearance, but with a straightforward, rustic taste at the same time.
I also got some cold Sichuan noodles. A classic to try at any Sichuan place. The noodles here were very good.
I also had a slightly different kind of cold jelly than I had the last time. This one had the distinction of having a huge amount of raw garlic on top. The spiciness, numbingness, and bite from the garlic was quite exciting to my palate, but this is only recommended to those who can stomach so much garlic.
Finally, there was the cold, spicy bamboo, which had the familiar spicy taste that all these dishes had, plus a refreshing texture.
About once every three days, I get a slew of comment spam. I've dealt with this in previous posts, and I've gotten MT-Blacklist to deal with the problem. My blacklist filter is huge, but still not complete enough to filter every spam. The filters that come from the master blacklist (no longer in existence) are very specific, and I find myself amazed that simple spam-related words such as "casino" and "porn" are not filtered out. So I ban these words outright. However, that means that no one can post a comment with those words in it. That sucks (although, for those words, no big loss). Plus, it's a stupid solution because spammers can easily work around it, substituting an endless variety of look-alike words like "casin0".
There is a solution, upgrading to Movable Type 3. However, since I have multiple authors, such an upgrade would cost around $40.
A coworker suggested I move my blog to LiveJournal, and let them take care of the spam problem. However, to use my current domain name, and to customize the look and feel to get my LiveJournal site close to my current design costs $20/year. Not bad, perhaps a better deal than Movable Type. However, I then have no control at all. If the spam problem (or other problem) still exists, I'm screwed. Also, I don't want to be part of the LiveJournal community, I'd rather be a part of the independent blogger community.
These solutions are not a terrible amount of money, but I'd like to see if I can make some quick fixes that can solve the problem for only a small loss of free time.
So, I have two potential solutions, one of which I already implemented. First, close off comments on most posts, which I did this morning. There is no easy way to do this through the Movable Type UI, so I had to just modify the database directly. Not too hard.
If that still doesn't stop the spam, I can do something quite simple like put a hidden field in the comment submission page, and look for that in the script where comments are posted to. That would thwart any tool that just does POST's to comment scripts. Alternatively, I could require that users enter some simple field manually to post. In that way, even if the spammers try to mimick a real posting, they will fail. It is easy enough to work around, but my site is not important enough for spammers to bother creating special code to deal with.
If all this fails, I'll have to cough up some cash and upgrade MT or move to LiveJournal.
The funny part is that these spams are useless! Movable Type changed the way comment url's linked, so that URL's in comments cannot effect Google page-ranking. But, spamming is so low-cost that even if 99% of the spam does nothing, it still is worthwhile.
Last week a few coworkers and I ate lunch at Fleur de Sel, a well-regarded modern French restaurant. They have a $25 prix-fix lunch. The lunch was three courses, the first was a frothy parsnip soup, with the cutest little parsnip raviolis you ever saw. Really good. Next I had duck & duck confit combination, which was an interesting contrast. The duck was perfectly done, and the light sauce complimented it nicely, without overwhelming the taste. Finally, I had a trio of fruit sorbets that had a really nice natural flavor to it. I didn't have anything that totally blew me away, but the parsnip soup came fairly close.
I'm getting to be more and more a fan of the prix-fix lunch. You can have a really nice, elegant three-course meal for a reasonable price. Yes, it is expensive, but a few cheap meals at other times will more than make up for it.
Another one I had recently was at Nino's Tuscany, where I had probably the best cooked pasta I've ever had. It was perfectly al-dente, which is really tricky to do and requires precise timing.
And, is it prix-fix or prix-fixe? I see both spellings.