Science Addiction

A dormant blog by Devanshu Mehta

On Comcast’s New Network Management Policies


Comcast has come a long way. Until recently, they were a company that deliberately injected spurious packets to slow down particular applications. They called that normal network management.

Now, comes an official Terms of Service and FAQ that address how Comcast will deal with network management going forward, and they are a huge improvement. Here are some of the key points:

Comcast has legitimate reasons to inspect network activity.
Let’s get the silly stuff out of the way first. Comcast has the right to do the following:

If we didn’t manage our network, our customers would be subject to the negative effects of spam, viruses, security attacks, network congestion, and other risks and degradations of the service. By engaging in reasonable and responsible network management, Comcast can deliver the best possible broadband Internet experience to all of its customers.

The question is, how do they implement it and how do they discriminate?

So how does it work?
This is pretty good- it is closer to a stupid network. If there is no network congestion, no one is affected. During congestion, bandwidth is shared as fairly as possible- except for users who are using the greatest amount of bandwidth. These users will get “managed” temporarily. This management will persist only as long as congestion exists, and will be based on current usage, not monthly aggregates.

How does it discriminate?
Based on high usage during congestion.

It does not discriminate based on protocols, applications, services or content. Any high bandwidth users will be affected, including ones using Comcast’s own streaming services.

Will it target P2P?
No, it is protocol-agnostic. If a P2P user happens to be a (recent) high bandwidth user during congestion, they will be affected. But not because they are using P2P.

Will it target competitors?
No. In fact, Comcast admits it will even affect people using Comcast’s own services.

What about the future?
They say they will participate in IETF and other standards bodies to work out reasonable, standard network management practices. And they will be transparent about new policies through the terms of service.

All in all, it is a tremendous step forward. Thank you to everyone who made enough noise to make this happen- especially Rob Topolski and the EFF. Also, thanks to Comcast for bungling this thing so bad that it got everyone’s attention and then setting a good example for other ISPs with these terms of service.

Rate the Debates

The conventional wisdom in the media about Presidential debate performances forms fairly quickly- within minutes of the debate having ended. Usually this is fed by only one measure- expectations.

For example, the media expectations for this Friday’s debate (if McCain un-suspends his campaign) are that Obama is not a strong debater and McCain does well. If these expectations are met or exceeded, it’s a win for McCain. If Obama is okay, it’s a win for him. Obama only needs to beat expectations, not actually be the better debater. The bar is higher for McCain.

The public has no way to affect the discourse on the debate. While the public are still gathering their thoughts after the debate, the cable news media starts pounding them with opinions and the conventional wisdom is formed by a few loud news analysts.

Rate the DebatesFreePress– a national, non-partisan, non-profit working to reform the media- has started

Sign up to Rate the Debates yourself and return here ( when they air to begin rating. We’ll tally your response along with thousands of others and inject our people-powered feedback into the news cycle – before the mainstream media pundits and spin doctors (mis)interpret the event.

Technology has given us the opportunity  to immediately insert ourselves in to the national discourse- but, for now, we need numbers to be taken seriously.

Great Schlep? Maybe I should d…

Great Schlep? Maybe I should do something similar for the Indian American community:

Radiohead Releases “In Rainbows” for Remixing

The band Radiohead is at it again. In a novel move earlier this year, they put their album In Rainbows on their site for download on a “pay what you want” basis.

They’re not done shaking up the way music is delivered to the people. From TUAW, we have this:

This week, they’ve done the same thing with the song “Reckoner” [iTunes link]. The six-track, DRM-free album costs $0.99US. Tracks cannot be purchased individually.

If you purchase the stems within the first two weeks of availability, you’ll receive an access code to a full GarageBand version of the song. When you’re done, you can upload your masterpiece here.

This is a fantastic idea to keep the fans involved and to keep the marketing buzz going. They already did something similar in the past with their song Nude, as did the band Nine Inch Nails earlier this year.

Political Polls Without “Cell-Phone Only”s


Mark Blumenthal of summarizes a recent Pew study:

Today they released a new must-read report summarizing findings from “three major election surveys [conducted] with both cell phone and landline samples since the conclusion of the primaries.” The verdict? “Pew’s surveys this year suggest at least the possibility of a small bias in landline surveys.”

Skeptics don’t think that missing “cell-phone only” voters is a big deal for a variety of reasons. Weighting a poll by age demographics should already account for any missed voters, but that assumes that those with landlines vote similarly to those without. The average “cell-phone only” user tends to be younger and hence- presumably- an Obama voter.

A few days ago Nate Silver at the awesome compared results from pollsters that include cell phones with a control group that does not and observed an approximately 2% bias against Obama in polls that did not include cell phones. The Pew study finds a similar effect.

Many people had predicted a similar effect in the 2004 election, but Bush’s margins over Kerry in the polls proved to be correct. Of course, there are many more people sans landlines in 2008 than there were four years ago.