BP's Weblog

To follow up on my last post about my VoIP experiences, I wanted to lay out the system I've built so far, and what it gets me.

I put together several of the tools I mentioned in the last post to make a system to allow family members to reach each other more easily. This is essentially already a fall-back system for everyone, since everyone in my family has a cell phone, and, consequently, calling on the evenings and weekends is already “free”.

I built a system where I tell everyone the toll free number of the system, and I give each of them an “extension” based on their names. It's pretty easy to guess how to reach everyone, so the only number that has to get written down is the toll free number, itself.

Once into the system, if you enter someone's name (as numbers - ie, 27926 is Bryan), you get prompted for what means they can be reached by. Normally, 1 for home, 2 for cell, 3 for work. You choose one, and the system puts you through. The called party sees your original caller ID, and you talk just like normal.

Why is this so great? Well, it works not only from home phones, but anywhere. And it's darn cheap. The cost of calling anyone in the Bay Area (me, my girlfriend, my sister, or her fiance) is only the cost of the toll-free leg of the call: $0.02/minute. If you want to reach anyone else, it's still only another $0.013/minute, for a grand total of $0.033/minute. That's with no monthly minimums or maximums, and through a toll-free number.

At those rates, I can give my family calling to each other “as a present”, and pay for it virtually entirely out of the income from the ads for other VoIP services that Google sticks on most of my posts about this.

Speaking of which - if you're not interested in how I do this, why not click on one of the ads and find out about some other service? I should warn you, at this point, most of the other services which show up as ads here charge monthly fees, assuming that something like $20/month unlimited will lure people in. And, it does, because normal phone lines cost just under $20/month without long distance and extra features. Rediculous - anyone I know who wants to can spend $70, buy a Sipura SPA-1001, and I will hook them into this system. The marginal cost per user is nothing. You can buy your own long distance minutes, have your own free incoming number, and dump your home phone line. At $16/month after taxes and fees, it doesn't take that many months to break even. At $0.013/minute for outgoing, free in-system (and various other destinations that are VoIP-enabled) calling, it's hard to imagine why this wouldn't be worth it.

Update: But why get help from me, when you could go it alone with one of the many services finally coming out to assist you in swapping your phone line for others. In particular, I just noticed the Bellster announcement, and there's been a technology called DUNDi for a while.

Posted by Bryan at 07:09 PM | Permalink | Comments (2)

So, I was reading this entry about Google's Local search. The example has you search Palo Alto for Cheap Restaurants. Try it yourself.

The top hit is a restaurant called Bistro Elan.... which anyone who's ever been to would certainly not categorize as “cheap”. It's not out of this world, either, but I have been there before with a group of 8, who were thankfully spending someone else's money: the bill hit nearly $600.

So, Google's Local search might still have a way to go....

Posted by Bryan at 08:30 AM | Permalink | Comments (0)

When I first caught this article on V2G (Vehicle-to-Grid power supply) I wasn't sure.

Let's see, sucking energy from the batteries of hybrids, which are just going to have to generate it back by burning gas. For peak demands, I guess that might make sense, depending on the cost/environmental impact of building to meet such peek demands. But the article suggests sucking power back from all-electric cars. Uh, what? The power company surely can build more efficient storage means than showing the power down the lines and back (once to charge the car, once to return it to the network from the car) on-site, with batteries that probably scale better. Not to mention that it supposedly costs hundreds of dollars to equip a car with the technology to do this.

OTOH, whenever the use of fuel-cell cars becomes widespread, this might make sense. I have a feeling that home generators based on fuel-cells will actually eviscerate any demand-cycle problems far better than building cars (and parking lots) that support plugging back in. Bah.

Posted by Bryan at 12:09 PM | Permalink | Comments (0)

To follow up on previous posts about Gmail and how I think that the "y" key, aka "Archive" really means the same thing that "delete" did in m previous mail system - Gmail's help center seems to have clarified. From the Gmail Help Center post on what the keyboard shortcuts are, it now specifies:

So, they admit that "y" means a different "archive" than the normal "archive" option from the Gmail drop down menu. Good to know that it unmarks Spam, though, I hadn't discovered that little gem...

Posted by Bryan at 11:28 AM | Permalink | Comments (0)

So, with the recent increases in Gmail invites, I decided it would be fun to experiment with eBay and the (expected) drop in Gmail account invite prices as supply flooded the market.

Here's the trend that I got:

If I had any invites prior, I'd expect to have seen the high end be closer to the ~$80 that I saw around June 1, but this is only my data, so I'm only reporting that.

Likewise, the number of Gmail auctions went from hundreds at the beginning of the month, to 4000-5000 open auctions over this weekend.

Worth noting: It costs ~$0.50 to open an auction of this type on eBay. So, closing prices over the weekend were dropping below break-even for most people tinkering in the market. In other words - gmail accounts really should be free, unless sold as larger lots. Unsurprisingly, there are several of these "5 gmail invites" or "10 gmail invites" kinds of auctions running right now.

And, yes, I have a couple left, which I will happily dispense to people that I know. If you know me, write me (here). They seem to keep topping off my invite balance anyway...

Posted by Bryan at 12:39 PM | Permalink | Comments (0)

Sounds like I need to better defend my position that "archive" = "delete" on Gmail.

First of all, the full translation of what I meant to say, but was apparently not getting across:

"archive" on Gmail is the same as "delete" on pre-Gmail e-mail systems.

At least, that's how it is for me. I've got nearly every message I've received in my personal e-mail since the fall of 1996. A simple procmail rule has created a monthly backup of inbound mail through my Linux server ever since I discovered I could. Even in 1996, disk space was so cheap that it didn't make sense to throw mail away. This has saved my butt dozens of times, when finding proof that someone made a promise, serial numbers of web-registered software, or just, generally, e-mail that I deleted.

The workflow was simple. My inbox was mail that I still cared to read, or remember I had to react to. A sort of todo-list of active discussions and items. When I was done with something, I deleted it from my inbox, totally assured that I still had a copy of it in my backup folders. If the message clung to some sort of theme or contact thread (in my case, groups by friend/family/work, and subgroups for each meme or person, depending on how general the discussion was), then I'd file the messages appropriately instead of deleting them. One of my coworkers has nearly every message she's received in 3.5 years of working here in her Inbox, for the opposite reason. If she deletes something, she might not be able to find it when she needs it. I just choose to use the "out of sight, out of mind" approach. She uses search a lot more than I do, I suspect.

(For those comparing how useful Gmail will be, my mail archives currently occupy 1428megs of disk space, of which 812 is the inbound stuff, and 6 are sent mail. Or, in other words, I'd have about 818megs of my Gmail account in use, had I gotten it in 1996. So, I expect Gmail will have a 8-15 year lifetime for me, at its current offered capacity).

Alright, so, why'd I say "archive"="delete"? 'cause that's what I do right now, and that's exactly the translation that is happening as I use Gmail more and more. When I'm through with a Gmail conversation, I hit "archive", just like the handy little gmail getting started guide tells you to. It just so happens, and the reason I was commenting, was that the "y" key, an apparently overloaded "archive/remove label" key, does exactly this, modulo the ability to apply several labels to a message. In effect, it does exactly the same thing as delete does on my current folders, assuming I ever copied a message to multiple folders, instead of moving it to one specific one.

So, yes, "archive"="delete". Gmail doesn't want you to ever delete a message, so you're supposed to shift your mindset to "archiving" mail that you want out of your attention threshold. If I'm in a label's conversation list, probably populated mostly by a Gmail filter, and I hit "y" while reading a conversation, it gets "deleted". Just like my existing e-mail system, it's not gone, merely forgotten. That's the way Gmail's designers assume you're going to use it, and that's exactly my point.

I've been using an inferior pre-pre-beta of a Gmail system for 8 years, and I have to retrain myself to use the beta release.

Posted by Bryan at 02:23 AM | Permalink | Comments (8)

Then perhaps you might be willing to contribute to FreeCache. This is kind of like an "Akamai" for the rest of us... and perhaps a little of what I've been talking about doing for a while among friends to distribute the load of large, fairly-static content from things like digital camera pics.

In theory, this does all of the things a solution I was talking about would do - reduces the burden on the source host quite a bit, provides replication outside of the burdened host in response to upswings in demand, and doesn't require too much centralized infrastructure. And users are none-the-wiser. I'm going to set up a proxy and see how it goes.

Too bad the file minimum limit is 5megs. This makes sense. Now, someone's just gotta' make a front-end Apache module which will redirect direct requests for too-large content away to freecache, except requests coming from freecache proxies. This should be pretty easy... anyone?

Posted by Bryan at 04:43 PM | Permalink

I think I remember seeing a talk he gave (probably a Forum) at PARC in 2002, but Dan Russell returned today to talk for the ISTL weekly lab meeting.

Here is a link to the notes

Posted by Bryan at 03:58 PM | Permalink

Seems like, living in California, sooner or later you start to think about all of the various environmental problems we're so attuned to here. My latest one has been quick/environmentally friendly commuting.

I haven't finished doing my research, but there are several useful sites around. A great starting place is: Electric bikes

There are getting to be more options in this space. Specifically, my favorite of the bunch is the ego-2, more of an electric moped than anything. $0.10/charge, they say, and, with after-market 36v batteries, even a big guy like myself should be able to take it up to 30mph. That's quite a decent commute time.

Posted by Bryan at 05:41 PM | Permalink

So, my girlfriend, let's call her J, is a grad student in genetics. We regularly get into discussions about what the pace of technology in her field is like... mostly because I work in high tech, and Moore's law is such a dominate force in my field.

So, the question is, how long until technology gets to the point where you can go to a clinic or doctors office, and, for a reasonable price, get your personal genome sequenced.

Another direction, which might possibly be attainable more quickly, would be automated phylogeny techniques. Phylogeny, for those not familair with the word, is the study of the ancestral history of an organism. Why would this be easier than doing a full sequence? I'd imagine it's possible to do differential analysis chemically, and isolate the smaller regions requiring sequencing.

What do you think? Should I post a longbet? Would anyone take it?

Posted by Bryan at 10:07 AM | Permalink | Comments (0)