Internet & Search
Get More Effective with Email: use Smart Folders and Tags in Thunderbird
Email overload is common between who works in technology. Launching a new Startup with my co-founders located in other parts of the US exponentially increased the number of messages that I/we receive and send every day.
While our email is maintained on Google Apps servers, I still prefer to use Mozilla Thunderbird as my email client. Starting from version 3 they added two functionalities that can become pretty handy if you use email as your main medium of communication: Tags and Smart Folders.
You can define tags following the menu “Edit -> Preferences -> Display -> Tags”.
As you can see, I defined 3 tags:
- End of Day (red)
- Tomorrow Morning (orange)
- End of Week (light blue)
I mark all the emails that I send/receive and should get my attention at the end of this business day with the first tag, the one for tomorrow with the second, and the one to check on by the end of the week with the third tag. It’s easy to keep track of things using these tags: every night, before leaving the office, I check to have handled everything I have to, and I do the same when I come in in the morning, or on Friday evening.
Smart Folders are basically “virtual folders” which show messages matching certain search patterns.
I created one that I called “Important Emails” and set it to monitor my Inbox for
- Unread Emails
- Starred Emails
- Emails with Tags
In this way I can just keep that folder open in my Thunderbird and be able to monitor all the important stuff going on.
Google/Bing, it’s Time to Use Classification to Improve Results Pages
I have been saying “users don’t think, they google” for a long time now, and every day I find new examples that confirm my theory.
Since users are this way, I wonder why Google and Bing aren’t showing many steps towards this. Sure, they inject news and images into the results here and there, but you can do it much better and in a smarter way.
I was discussing of this with my friend and ex-colleague Francesco over Christmas, but the idea goes back 2~4 years with my discussions at Ask.com with Apostolos Gerasoulis and Antonio Gulli.
If you have tons of traffic, and some sort of content type and category classification, then learn from the clicks of the users how to organize the page. For example, if 70% of the people looking for “apache” still click on some links about the Indians, then 7 out of 10 links should be picked from the same category, and we should stop showing 9 links about the webserver (come on, afterall every tech guy will click on the one of apache.org anyway). How to do this? Categorize the query based on what people click on, and then boost results from the same category.
Same applies for content type. bet everyone who looks for “Justin Bieber pictures” on Google would love to be brought to the image search page directly instead of the normal one plus some pictures injected. If almost everyone who search for the title of a song then click on the YouTube video on the top, don’t you think you can offer a better experience? Perhaps expand that at the top automatically, so I can just hit play.
One could go in more deep personalization (e.g., I may be in “work mood” or at home), allow different profiles for each computer (e.g., me, my wife, my kid, …), learn what I like from my choices, etc, but already offering a standard set based on the average (i.e., 99%) user could go a long way.
Sure, some purists will be annoyed by that. For them you put a link somewhere which sets a cookie and avoids to do this. But for the rest of them, which is probably 99% of the Internet users of today, this would be great.
People don’t think, they Google. So, make it easier for them.
Twitter “Similar To” allows you to Discover People with Matching Interests
I just realized that Twitter now has a “Similar To” panel in the profiles pages of the users. I am not sure how they compute those but it is a pretty cool feature, which allows you to expand the people you “follow” beyond the most popular ones.
If I had to build that service (or a competing one), I would probably use as similarity factors:
- a signature of the most discriminant terms that each person uses in their tweets
- the links they posts (or at least the category)
- the social network (who they follow).
I wonder if that is how they did it.
Use Skype-To-Go to Forward Google Voice Calls to an International Number
I like Google Voice a lot, especially when I am outside of the US. With any data-plan, I can see text messages and listen to voicemail left for me to my US number. Since the international call fares offered are reasonable, I would love to transfer the calls to the international number I am using in that moment. Unfortunately, that is not possible at the moment.
But here is a simple work-around. Log into your Skype account and get a Skype To Go number, then register your international number as one of your favorites.
Once you have registered the Google Voice number with Skype, you will receive a direct number to call for each of your favorites.
Save that as forwarding line in Google Voice and forwards all the calls to it.
Done!
In case Skype does not really let you register the phone with them, just install the Google Voice application on your phone and/or call the Skype-To-Go number from your gMail. It will say that it does not recognize the number and make you go through a 2-steps verification process. It is relatively simple and in a few seconds you will be finally registered.
Here is Why You Should Not Try to Build a New Search Engine
If you are thinking to build a new search engine, stop now. If you endure, you will be able to exploit domain authority and relevance, but likely not much else. In the end, you will build something close to the old Lycos or Altavista, which can not compete against big giants like Google or Bing.
It is not by accident that other pioneers (e.g., Ask.com and Yahoo!) of the search engine world shut their door recently, and the explanation is simple: they did not have enough traffic. The traffic I am talking about is not the one that generates revenue through ADs impression. I am talking about real user traffic, made out of queries, clicks and analytics.
In the last years the topology of the Internet changed a lot. Long gone are the days in which it was a fairly connected graph, with every Geocities user creating its own favourite links list. Today the large majority of the links that you can find are either self-referential (e.g., past articles in the same blog/site) or fake (e.g., trying to push the ranking of some site).
For most the new users of the Internet, content discovery is mainly done through Facebook or Twitter. You log into your account, and take a look at what your friends posted: the last video of Lady Gaga, some funny flash-based game, or some interesting piece of news. But even if you find it interesting, you will not add it to your website (assuming you even have one). At most, you will share it with your other friends on Twitter or Facebook.
On the other hand, when you know what you are looking for, you search for it on Google and click on a bunch of links. The large majority of the content that Internet users access today is at distance 2 from them, and if they happen to click on some more links, it is usually just because they are navigating within the same website.
Links were the fundamental block of the first Google ranking algorithm (PageRank), which allowed the search engine to beat Altavista or Lycos, which were still just based on text relevance and domain authority. It made sense: if you had lots of external (i.e., not from your own site) and important (e.g., from the Yahoo! homepage) pages linking to yours, you must have been doing something right, and the ranking of your site would increase.
Now that links are scarce, search engines must rely heavily on user behaviour, analysing the clicks and the time spent on the pages. After each search, a user is presented 10 links in the result page, some directly visible and others hidden by the screen size. The links at the top were ranked higher by the search engine, but if a user ignores them and clicks on the 4th link, it is giving away a very important piece of information: it does not agree with the results ranking.
Similarly, if the user clicks on a link, but then a few seconds later hits “back” and returns to the result page to pick another one, it probably made a bad choice or the page was not interesting. Repeat this process for millions of queries per day and you have a great opportunity for learning and improve your ranking algorithm.
Google and Bing serve hundreds of millions of searches per day and this puts them in an unique position to take advantage of user behaviour. On a small portion of their searches they show new URLs (even if there are no links, e.g., a new YouTube video) and look at the behaviour of the user. If users seem to appreciate, they incrementally show it to a larger audience until they determine the correct ranking.
Any new search engine does not have this luxury. It will start with no traffic, or at most with internal one generate by the employees and perhaps some editors. With no links, they will have to rely on relevance and domain authority, or at most classify/tag the pages trying to more closely match the interest of the user (e.g., are you looking for “Apache” the search engine or the Native Indian Tribe?), and this will allow them to get only that far, assuming that their technology can even keep up with the daily growth rate of the Internet.
While things are hard for new search engines, Facebook has a real shot at Google. Its 600 Million users spend on average about an hour per day on the site, chatting with their friends, browsing their friends’ pages, clicking on videos or links posted on their walls or messaging each other. That’s a lot of information. Facebook is in the unique position of creating the next generation of search engine, which take advantage of your location, your friends and your interests to personalise the ranking of the pages.
Looking for a pizza? They can use your location to pull local listings, then check how many of your close friends (the ones you talk the most with, for example) “like” those places/links, and finally weight in the votes of other users in your area. For more general searches they can use the interest in your profile, and the ones of your close friends, to target the results. They can evaluate how many people clicks on each link, where they come from, and if they are similar to you (e.g., comparing the profiles).
Facebook is in a very unique position. People are afraid of Google spying on their wifi but not of sharing interests, relationships, pictures, phone numbers or addresses with Facebook. Microsoft invested in Facebook a few years ago and they still own a percentage of the company. If they play their cards right, they can very-well co-own the search business.
