Sunday, May 31, 2009

Non-scientific first impressions of Bing

So Bing is out, so I decided to give it a quick try before I quit for the evening. So I ran the last 3 searches that I did (only one of them was actually an all-web search, but I ran all of them anyway), and the results were quite interesting. They reminded me of the company that is behind this:

1. scalaz (the only web search): Bing thought I misspelled "scalar" and returned only "scalar" query results. There was a small link that I could press to return only "scalaz" results, which when I clicked worked ok.
2. java posse: The first hit was the website I was looking for. However, when you look at the "special internal links" you see "January", "February", etc. What are they? Links to some months of the show (it's actually a blog), but it doesn't contain the year, so not that useful. Also it doesn't contain all the months, just a random select few.
3. snooth cork'd: Again it tried to rewrite my query and use "smooth cork'd", which was quite odd. What I was trying to get is a traffic comparison between both wine organizer/social network sites. The best answer that I found? Wolfram|Alpha!

As you can see, there is an underlying common pattern here: more than Google, Bing tries to predict what you really want and, consequently, will get more things wrong. That's why Microsoft Office is so annoying sometimes. Maybe it's the same engineers that now moved to Bing! :-)

Some other day I'll spend more time on this analysis and find Bing's strengths.

Saturday, May 30, 2009

I love website bugs!

Sorry, a little software development joy. I was browsing around food links that I read on this article and when looking at ethnicgrocer.com and decided to look at what they think French Pantry Essentials should look like, I received the following message:

Server Error in '/' Application.
(...)
Parser Error Message: The server block is not well formed.
(...)
Line 7: <%@ userAgent = Request.ServerVariables("HTTP_USER_AGENT") userAgent = lcase(userAgent) if Instr(userAgent, "googlebot") then Response.redirect("http://www.myethnicworld.com/") %>

You can see the full page here. Look also at the source code. In the bottom you will see the stack trace and a very useful warning message:

<!--
This error page might contain sensitive information because ASP.NET is configured to show verbose error messages using <customErrors mode="Off"/>. Consider using <customErrors mode="On"/> or <customErrors mode="RemoteOnly"/> in production environments.-->

Should I be the person seeing this? I really don't know why this is still the default behavior on web servers. It should be aware of the client and just dump debugging information if you have some sort of cookie set. Enough of seeing other people's code out there!

Friday, May 29, 2009

Repeating today's buzzwords

Today is the day of two new big service promises, but no deliveries: Microsoft's Bing and Google's Wave. Bing will apparently release next week for people to see, so maybe it's a better promise. Wave is supposed to only release later in the year, so I have to control my skepticism. But, on the other hand, the Wave presentation had so much more information than the Bing presentation that it's easier to be more excited about the former.

What are my personal impressions? Wave looks confusing, but maybe as an engine it might work. I'm concerned about the use of a "fake participant" on a conversation to do integration with external data sources. Sounds like a weird architecture (especially because it seems to work differently from a normal person). Identity is also a little interesting to discuss. But it's too early to tell.

Bing? That one is even harder to gauge. It seems like people that had access to it are excited about it. By only having access to a silly video and some screen shots, I can't see it. If Wave looked confusing, Bing looks even worse. The same interface for a lot of different use cases is generally a bad sign. I'll try to spend some time playing with it next week when it launches and then I'll give a real summary of my impressions.

I'll have to add that lately I have been trying to reduce my reliance on web search and I'm trying to focus on tools that are better for what I'm looking for. For example, if I'm looking for a website, I go to delicious or Twine. If I'm looking for a product, I first look for review sites and then the product. If I'm looking for a person or place, I go to Wikipedia.

But there are a few use cases I still need to fall back to web search. This is mainly important at work. For example, code samples, documentation, error codes, all of them are generally very hard to start with a specialized web site. Another annoying thing is that the integration of web search in my browser is easy. All other integrations are a little clunkier, which makes me sometimes use web search simply due to convenience.

Anyway, I'll continue keeping this goal in mind and see what it teaches me.

Sunday, May 17, 2009

Web 2.0 maps... Should we really care?

I came across this on FriendFeed today:

From Dot.com to Dot.gone - Web 2.0 Start-ups That Have Vanished

An image with Web 2.0 company logos that were active in 2006 and then annotated with which have gone out of business and which have been bought by other companies. In other words, it actually has one case of success (being acquired) and two of failure (going out of business and still being around - ok, some do have this odd plan of making money and being around, but this map doesn't identify the ones that are actually profitable right now).

I thought that the most interesting and telling thing of this map is that I probably only recognize about a quarter of them. Certainly I'm not in the target audience for most of those websites, but 25% seems a little low for companies that are hoping to get somewhere.

Wednesday, May 13, 2009

Shorts for the day

Life has been pretty tiring as of late. There was a big launch on Monday evening and that made me stay up pretty late and last night when I finally was able to sleep some I had dreams about co-workers mocking me for not hiring some candidate. Quite a joy.

Launch, you say? What launch? I didn't see anything different! Well, that's how it works for us back-end people. And this was one of our most visible launches, so it's out there somewhere. Just don't get confused by it.

Another thing that I wanted to mention is that I've been reading Halting State by Charles Stross and there was one thing that I found funny. At some part of the book they were trying to track transactions that happened at a specific time to see what happened with something and one of the characters say: "Good luck grepping through those logs. It's almost a million entries a minute!" And they promptly drop the plan. A million log entries a minute for a bounded time range? That's really not that hard... Or maybe I'm just spoiled.

Sunday, May 10, 2009

Back from Mount Si

Today I finally went to what is supposed to be Seattle's most popular hike, Mount Si. Unfortunately I think its popularity comes only from the proximity to Seattle (about 45-minute drive) and the good condition of the trail itself, but not really how nice the hike is. It's an 8-mile roundtrip hike where you hike up for 4 miles straight (gaining about 3500 feet, about 1 km) and then down for 4 miles back. There are pretty much no views in the middle of the way. If you don't make it to the top, you don't see anything beyond trees and some rare wildlife. But is the view from the top worth it? Oh, yes! It's amazing. If you get a clear day (which we didn't quite), you can see all the way to the Olympic Mountains.

Conclusion: did I enjoy it? Yes! Would I do it again? I don't think so.

Here are some pictures:

This is to show that unfortunately the day wasn't so clear. In this picture you can see the Olympic Mountains in the back and some hazy version of downtown Seattle and downtown Bellevue.



And this is the most interesting wildlife shot of the day (unfortunately not that sharp because it was a little dark from all the trees to take picture of a moving object):



If you can't identify what you are looking at, it's a snake, a Gartersnake to be more exact. Not dangerous at all.

Wednesday, May 06, 2009

RSS in the news

There were two interesting articles on RSS published on "almost mainstream" news:

Steve Gillmor's Rest in Peace, RSS

and

Dave Winer's reply RSS is dead? My ass...

The interesting thing is that they approach the subject on two different ways and, of course, get to two different conclusions. I'll let you read the articles and get your own conclusions.

I think both are right, and both are wrong at the same time. On one hand, what Steve Gillmor is saying that what RSS lacks is the community is right. RSS is a one-way street that only supports third-party distributed commenting (by using whatever your reader supports, like Google Reader that supports commenting, sharing with your friends, and flagging news as interesting). And that's certainly a problem, as you lose people adding to your article.

However, I don't think that centralizing everything around a twitter-like approach, or FriendFeed, actually solves the problem. Look at me right now, I'm technically commenting on two articles at the same time and adding my point of view (that has more than 140 characters). That doesn't fit any of the two options.

I really like the concept of real-time that a short messaging system provides, but I also think that there is a need for other streams of information to be added based on your experience of what you like to hear about. So I want both, but I want both to work together as one. And add Facebook to the mix so everything becomes complete. I think of it as multiple dimensions working together:

Friendship: I do care what my friends are up to and worrying about. I don't really matter to get noise from them like "I wished Illustrator would stop crashing". But if you are just a person that from time to time has interesting thoughts to share, I don't care about you getting a free KFC grilled chicken.
Internal growth: for the things I'm passionate about, I want to be able to get in-depth discussions about it. I care to know about what the people that I think have good ideas about the subject are talking about, and, as I find new sources of information, I want to be alerted when these sources provide more information.
Society need: We all need to know what is going on around us in the society. We shouldn't hide from it, just because it seems boring and always the same. If you think something is boring, maybe for those things you just want a count of how many people are talking about it. For new things, you probably want to follow the development of it. So the system would be something like "alert! There is something new going on with..." and then you can select to "follow" the news (and maybe post a question about details you think are missing in the news) or just say "whatever... good to know, but I don't care to get more details"

There is an interaction between everything. My friends might want to make sure I'm following a story that I might have thought it was not interesting enough for my free time and vice-versa. People from my interest group might migrate to becoming my friends. Things that are just news might become part of my "internal growth" interests.

The question that remains is how you can get to it in a distributed manner. Building one big silo that contains all the dimensions is probably not what we want. Depending on my interests, the features that I care about are different. For example, if I care about photography, a good photo cataloging and visualization is needed. If I care about electronics, I need to have access to schematics. If I care about shopping, it would be good to have links to places where I can buy the products or navigate through their features. I don't think anybody can build a website that can handle all those use cases and still make it understandable.

Alerts on news and filtering is also very different than friend activities. For friends you care about dates and locations, and passing ideas. For news you care about classification of alerts and identification of story development. And actually it also has some features of the "interest" case above as different types of news probably require different interfaces to be able to efficiently understand and act on this news.

So what are the underlying elements that need to be combined? I can think of two main things:

1) People: we need a FOAF-like solution to identify people and be able to relate them across pages.
2) Subjects/interests: for you to be able to correctly alert on the things that I care about, you need to understand that I care about that. And as you are providing your interests in a separate website, having those being shared will certainly help.

Anyway, what started with a short discussion about two blog posts, ended up becoming this long article that I probably should spend more time thinking deeper about it. Actually, on the contrary, it smells like another project for me and I don't want to have another one on my list of things to think about and never getting to it (I've been keeping those on Evernote and I've been pretty happy about the experience so far. I just wished they had a Blackberry app)

Tuesday, May 05, 2009

And what happened with the cheese?

I just realized I forgot to tell the end of the story of the cheese that I mentioned quite a while ago. So... It's actually done! How did it turn out, you might ask... Well here is a picture:



It feels a little bit like cheddar. It's a little crumblier than usual, and maybe a little on the sharp side. But nobody has fallen sick from eating it and I was happy. I'm not sure how long until I'll try it again, though. Probably I'll go for something different for some time, like brie, or a blue cheese. Then I might try another hard cheese. In any case, it's always very hard to make something that you'll only know if you were able to get it right many months from the day you did the work. At least hard cheese not like wine that you work for many days and then wait. You just have a few hours of work and then a few hours of cheese pressing and that's it. Blue cheeses require a little more constant attention.

But if I really want something that is quite risky to do at home, and requires a lot of attention, I should try to cure some meats... :-)