Barry Shein has had enough of spam. And he isn't going to take it any more.
Barry was at the spam conference at MIT last month.
How will people's behavior change as a result of spam? Will people learn how to write a message differently so that it will not get caught in a spam filter? Will people learn to use plus aliases? Will they learn to include special words in the message, such as the recipient's first and last name, so that the message will not be caught as a false positive? Will they come to understand that email is unreliable because of spam filters? If so, what will they do when a message is really important? How long should you wait to resend it if you don't get a response? Should you always respond to a message, so as to acknowledge you received it? If a message is really important, and you have more than one email address for a recipient, should you send it to all the addresses?
Another idea about filtering spam. To reduce the occurrence of false positives, search in the body of a message for certain words. If you find those words, declare the message to be non-spam. The key words could include names of your company's products, names of family members, organizations that you belong to, and so on. A good idea would be to include one or more unique words in your signature, and search for those words in incoming messages. If you find those words in an incoming message, then it is likely that that message is a reply to a message that you sent.
Okay, here's my idea for a spam filtering technique: Find the URLs in a message, then look up the domain names from those URLs in the WhoIs database. If the domain name was registered less than one month ago, then call the message spam. Similarly, if a URL has no domain name, then call it spam.
To improve this techinique, create a database of URL domain names or IP addresses that are used by known spammers. Perform look-ups on those names. That helps to cover the domain names that are used more than once, and which may have been registered more than one month ago.
Unbelievable! Trying to get $17,000,000,000 by beating a dead horse.
Dave Winer mentions OddPost.
I had never heard of OddPost. But after browsing around their site, I like the company. I like their distinctly non-corporate persona. No marketing mumbo jumbo. I would be much more likely sign up for a web mail account with OddPost than with Yahoo. Why? Because Yahoo is too corporate. There are many layers between you, the customer, and real people at Yahoo. Something else, too. I would be much more likely to tell my friends about OddPost. When are companies going to learn that it's okay to let down your guard? To let real people in their company talk directly to customers? To show some personality?
Welcome to Fredrik Lundh's effbot.org site.
I like it!
Joe Gregorio writes about The Well-Formed Web.
Joe's RESTLog looks pretty cool.
A Daniel Brandt complains about Google's Page Rank.
A complaint that the search engine's users don't get good results is valid. A complaint that a webmaster can't get his pages to rank highly for certain generic search terms is not valid. We have to take the search user's point of view, not the webmaster's.
We all know what search was like before Google: page after page of totally useless results. So even a complaint from the search user's viewpoint must take into consideration what is possible. Google is a remarkable improvement over the first generation of search engines.
Here's another interesting question to ponder: How did Google itself become a popular site? No, they didn't rely on search engines.
Where would we be if we didn't have free, throw-away email accounts available, like the accounts from HotMail and Yahoo?
Here's one thought: Without throw-away email accounts, people would be much angrier at email marketers, including the e-tailers who think they are legitimate. As it is, anyone who is not a novice at email knows to get a throw-away account when they need to communicate via email, but don't want to compromise their real email account to email marketers.
In a way, this is a really disfunctional relationship. There are so many people or companies out there that will abuse email, that email users are driven to sign up for throw-away accounts. The companies that offer the throw-away accounts know that these are throw-away accounts, which are likely to receive a much higher percentage of spam. But they offer them for free, probably because they think they can push advertising on their accounts' subscribers, or maybe because they want to be able upsell their subscribers to a non-free account.
Can this situation continue? Can Yahoo, Hotmail, and other companies continue to offer free, throw-away accounts that get a disproportionately large amount of spam? Does the current situation change if the spam problem gets worse? What happens if we get to the point where free, throw-away accounts are no longer available?
Google Named Brand of the Year. UPDATE: A poll by Interbrand finds Google the top global brand in 2002, beating out Apple and even Coke. [internetnews.com: Top News]
This is amazing. Consider that Google did not spend millions to build its brand via television advertising or other media. Google gained brand recognition primarily by word of mouth -- web users telling their friends about it. There's a lesson to be learned here for anyone who is in business.
Russell Beattie: Thoughts on Open Sourcing Your Code: The OSS Prototype License
Spammers know my email address, but none of them seem to know my first and last name. I'm sure it's not as easy for them to get a person's first and last name as it is to get the email address. Perhaps a very good way to reduce some of the false positives in spam filtering tools is to look for your own name. Perhaps a good way to reduce the chance that your own message will not trigger a false positive is to use the recipient's personal name. A good rule might be to look for the personal name on the first non-blank line of the message, as well as in the To or CC header fields.
Comcast Internet: Effective November 15, 2002, Comcast has revised the Comcast High-Speed Internet Service Subscriber Agreement by removing the virtual private network (VPN) restriction. Previously, Comcast only permitted the use of VPNs in connection with the Comcast Pro High-Speed Internet Service. Comcast has also revised the Subscriber Agreement to clarify that certain Service plans may permit or recommend uses of the Service not otherwise provided for in the agreement.
AOL became the most successful online company in the world. (Indeed some would call AOL an ISP, but that's debatable, and depends on one's defintion of ISP.) They became successful by taking a new technology -- better, the "Next Big Thing" -- the Internet, and making it available to ordinary people by making it simple to use.
Now here's an interesting thought. Will a company, such as Lindows or Red Hat, become very successful by making Linux available to ordinary people? I understand that there are lots of differences between the Internet and Linux, starting with the argument of whether or not Linux can be considered a "Next Big Thing" at all.
Robert Scoble has a very good entry in his web log: Microsoft is jealous (and scared of) of Slash Dot and Scripting News.
I certainly have noticed Microsoft's difficulty in creating community.
John Udell: Let's promote scripting languages to the status they deserve.
Is there a word for that place in an online article where you can scroll far enough to hide the annoying animated ad?
Why do companies put their DLLs in the Windows system directory?
What will happen when applications written to run on the .NET runtime are common? Will companies insist on putting their assemblies into the GAC? My guess is that they will, although I don't know why for sure.
But what about security. How secure is .NET?. More specifically, how secure are applications written to run on the .NET runtime? Certainly the "binaries" in .NET are easier to disassemble than native machine-code binaries. I'm not the expert here, but I assume that means they are much easier to infect. Assemblies installed in the GAC can only be tampered with by an administrator. Does this mean very many assemblies will be installed into the GAC?
I understand that there is code signing. But it seems to me that a virus writer, if he targets a particular application, could just copy the disassembled code from that application, and then create a replica that is unsigned.
Update: Microsoft says 'Open source threatens our business'
Almost every software company must come to grips with open source software.
Whatever happened to Watcom C++? Obviously, it couldn't compete against Microsoft Visual C++. Well, I found it. It's being made into an open source product, and it's being ported to Linux. It's nice when software turns open source instead of into abandonware.
I'm just looking at the Open Source Initiative approved licenses. I see an interesting comment that, while translations into other languages are available for some of the licenses, the English versions are the only ones that are legally binding. That has me thinking: assuming a common language is needed for worldwide legal contracts such as software licenses, will English become the de facto language for such contracts? Is a common language even necessary?
I spent Monday optimizing some C++ code that parses MIME messages. My first benchmark surprised me -- the parser was unbelievably slow. On the Sun workstation, I used the profiling tools to analyze the performance. Then I began making various tweaks wherever I found a hot spot. In the end, I achieved a speed-up of 70 times. There are two lessons to be learned from this exercise. First, it is possible to write very inefficient code in C++. Second, profiling tools are extremely valuable in helping one to increase the performance.
On Tuesday I ran a comparison test of the C++ code with similar Java code. The Java code ran roughly twice as fast as the C++ code. I imagine that I could continue to tweak the C++ code to get better performance. However, I expect continuing to tweak the C++ code would mean losing some of its functionality, or else making the code uglier.
Answering my earlier question: Perhaps it's possible to create web-based email that uses primarily stateless HTTP. I don't know enough about the options for authentication to know if the use of HTTP could truly be stateless. Would the server have to maintain state associated with the user authentication? Or, would that "state" be insignificant to the point that it's pretty much transparent to the developer/designer. For example, could each email account have its own directory that is password protected by the usual mechanism provided by the web server (.htaccess).
Which is more scalable, IMAP4 email or a web-based email?
At first one might think it has to be web-based email, because it uses a web-based architecture, which has proven scalability. If you think deeper, though, maybe that's not necessarily true. The web is highly scalable when you think of it in terms of a REST architecture, but that implies a stateless architecture -- HTTP without cookies, if you will. When you must maintain state, then the "web-based" architecture is not as easily scalable as a stateless architecture. Now, consider a connection-based protocol like IMAP4. Is it scalable? Well, a TCP connection requires the host to maintain state, namely, the state of the connection, identified by a quintuple (remote address, remote port, local address, local port, protocol). So, it's really not that different from a stateful HTTP protocol, except that in the case of IMAP4, the state is maintained by the operating system and in the case of stateful HTTP, the state is maintained by the application. When HTTP needs to scale to hundreds of thousands of simulaneous session, a reverse proxy or some other device is required to distribute the load among mulitiple servers. Could one create a reverse proxy for IMAP4? I would think so.
In summary, I don't think developers/designers should be quick to write off connection-oriented protocols like IMAP4 in favor of HTTP-based protocols.
Spam deluge leads to search for silver bullet
"Spam is a huge concern for us because consumers are just erasing everything. They don't know the difference between spam and legitimate marketing," said Christina Duffney, a spokeswoman for the DMA.
Ms. Duffney, you're wrong. Consumers know the difference. It's the marketers who don't know the difference between spam and legitimate marketing.
Would someone please create an XML dialect for email content? Start with XHTML, for example. Drop some of the structural tags like h1, h2, etc. Include tags like font, big, small. The fact is, that email is not structured into sections. If email has any structure at all, it has sections like opening, body, closing. Style sheets with email is way out. Who has time to thing about styles when dashing off a quick email?
From Doc Searls: The Net was built for fat, symmetrical, end-to-end sharing of everything, with no value-adding intermediator in the middle. It wasn't built so big dumb companies could use it as a one-way sluice for their own "content." Yeah, the Net'll support that, but that's not what users want it for.
Broadband ISPs like Comcast, according to their terms of service, forbid subscribers from running "servers". I understand that subscribers who run "servers" could be taking advantage of the Internet service -- for example, to run a web server for their company -- which could mean that they grab a disproportionate share of shared resources. However, what's so terrible about a subscriber running a web server so that their relatives and friends can view the pictures they took at a birthday party? Or so that the subscriber can check his mail while away from his home?
Do the broadband service providers get it?