The world according to David Graham
Displaying the most recent stories under internet...
A podcast with Michael Geist on technology and politics
Over the few years I had the responsibility of representing Laurentides--Labelle in Parliament, I spent a great deal of time and effort talking about technology and their related issues within politics.
One of the people I had the opportunity to meet along the way is Professor Michael Geist, Canada Research Chair in Internet and E-commerce Law at the University of Ottawa Faculty of Law, who I had been following for years.
After my defeat, he invited me to his office for a conversation about the experience of being a technologist in national politics, and you can listen to the conversation on his blog:
The LawBytes Podcast, Episode 32: Reflections from the Open Source Member of Parliament -- A Conversation with Ex-MP David Graham.
words - whole entry and permanent link. Posted at 16:21 on
November 18, 2019
Guelph finally on high-res Google satellite
A short note to point out that Guelph is finally represented on Google's high-res satellite map after spending quite a long time being an island of low-res completely surrounded by high-res. Lafarge property is visible, my house is visible, and judging by the state of my back yard and which cars were in the driveway, as well as the parking lot at the Via station and the construction of the new City Hall, these shots were probably taken on a working day in May of 2006.
words - whole entry and permanent link. Posted at 03:49 on
April 17, 2008
If you blog from or about Guelph, there's now a blog aggregator for you. GuelphBlogs.ca has just been launched to aggregate all of Guelph's various bloggers, regardless of political stripe or topic of discussion.
To get your blog on this blog roll, simply send me an email at email@example.com with the address of your blog and I'll check it out and probably add it to the database.
If you have any suggestions for improvements to the site, which I would like to keep ultra-low-maintenance, I'm all ears.
words - whole entry and permanent link. Posted at 15:55 on
March 24, 2008
New liblogs code update
New liblogs code went up last night, and if you're wondering what's changed, here goes...
Among the features:
- The entry link now goes to the blog entry, rather than to the blog. You can click on the blog name to go to the blog.
- The site is in a never-ending loop rather than on a timer looking for new blog entries. As it finds them, it posts them immediately.
- Timestamp cheating is no longer supported. An entry posted too far into the future will be totally ignored.
- Timestamp and entry revisionism is no longer supported.
- If your blog changes its name, the site will now pick up on this fact.
- If a site goes down, it will be automatically taken off the syndication list until it comes back up, when it will automatically reappear.
- The main blog index automatically refreshes itself every half hour, so you can leave the browser open and check back periodically.
- Beta Blogger is now supported.
If you have questions, comments, bug reports, or any other feedback, feel free to email me or post a comment.
words - whole entry and permanent link. Posted at 13:44 on
November 17, 2006
Wireless Internet access in Canada is behind the times
A few years ago, I bought a Fido GPRS (general packet radio service) card for my laptop, and an unlimited data plan. Two years ago, Rogers bought Fido, and now, the GPRS service is virtually unusable -- except in Ottawa.
When I first bought the GPRS card, the unlimited data plan worked without interruption pretty much anywhere. Using the dial-up speed cellular Internet connection, I could get on the Internet from anywhere within Fido's limited service area. In south-western Ontario, that was pretty much anywhere between Oshawa and Windsor. It also worked in much of the US. The GPRS card worked in Linux, a critical requirement for any laptop Internet adapter for me, and everything was good.
Two years ago, Rogers bought Fido. I was concerned that the company would be gutted, but was happy to hear that Fido would be a more or less independent company within Rogers, with its own phones and its own plans, and its existing customers would not, ostensibly, be affected.
Not long after that, my GPRS became less reliable. A year later, it became evident that Rogers was removing Fido's infrastructure and places where I used to have Fido service but my friends did not have Rogers service quickly disappeared.
My GPRS connection began to hang up on me. If I used the connection actively, it would hang up on me after 12 minutes, like clockwork. Because the IPs are dynamic, any time I reconnected, all my TCP connections were broken. For the non-technical reader, that means it became a pain in the rump to use. A call to Fido's once-useful tech support yielded no useful information.
I spent the past week in Ottawa attending a conference (more about that here: 1, 2, 3, 4). On the way there, my friends and I used the GPRS card to get on the Internet from highways 401 and 417. As always, the connection cut out every 12 minutes, like clockwork. As we approached Ottawa on the 417, the connection stayed up. It worked all the way to our hotel with no further interruptions.
A couple of evenings later, still at the conference, still in Ottawa, a dinner reception room had no wireless reception. I put my GPRS card in my laptop and got on line. It survived, uninterrupted, through the reception.
Yesterday, we left Ottawa, and again connected through GPRS. It required no reconnection and worked reliably until we got outside of the Ottawa area. The rest of the way home, it cut off every 12 minutes.
I can only assume that Rogers does not wish to irritate government officials from Canadian and former governments by hanging up on them, but us lowly regular customers are clearly not important enough to them for such consideration.
I'd like to get a Blackberry, as nearly everyone seems to have lately, but for something comparable to what GPRS+laptop gives me, the cost is utterly exorbitant. My Fido GPRS unlimited package has been grandfathered. It is no longer offered, but as long as I don't cancel it, I still have it. I'm therefore very reluctant to give it up without a better option.
Fido does not offer a Blackberry, though they do offer a competing device called a hiptop. Telus offers a CDMA-based blackberry with limited data and a TCP stack. The TCP stack works, but accounts I have read on line are that some ports are blocked. This means that the blackberry is capable of using the Internet like any connected computer, except that some services are blocked. Bell offers a great, high data limit plan on their blackberry, but offers no TCP stack. Rogers offers the best blackberry plan: a full, unblocked TCP stack -- but at a cost of around $100 per month with a data cap of a mere 25MB. I would use that up in a week without difficulty. At least they no longer seem to be calling this "unlimited".
I'm still holding out hope that we will have inexpensive, widespread, third-generation wireless Internet access in Canada before Europe and the far East arrive at the next generation of wireless connectivity.
Meanwhile, I'll carry on with the GPRS, reconnecting every 12 minutes.
words - whole entry and permanent link. Posted at 18:07 on
July 24, 2006
Patent application jeopardizes IETF syslog standard
The Internet Engineering Task Force is working on a proposed standard for the ageold but never standardized syslog protocol, but its efforts may be in jeopardy thanks to a patent application by Huawei Technologies Co., Ltd., of Shenzhen, China.
Syslog working group member Rainer Gerhards describes the purpose of the work this way:
Syslog is mostly a de facto standard. This means that no "real" (officially written) standard exists.
Syslog has a number of security shortcomings. An IETF working group was created to solve these security issues. A sideeffect of this is that the protocol itself must be standardized. One standard RFC 3195 is already published. RFC 3195 is a big departure from traditional syslog and not really accepted by the market. The IETF syslog working group is currently trying to standardize
a) the syslog message format
b) messages transmission over a secure transport, namely TLS
In practice, syslog is already being used over TLS (or SSL, which essentially is the same).
However, these solutions do not interoperate very well and the big players (like Cisco) do not natively support it because there is no official standard (at least this is my best guess why they do not).
In even shorter words, the IETF is trying to standardize a secure way of doing syslog. A way that is already in widespread use today.
Some members of the working group, including cochair David Harrington, work for Huawei Technologies, and the company has submitted a patent application. The contents of the patent application have not been released, but the IETF working group has been informed that their standard may conflict with the pending patent.
Huawei's proposed patent licensing declaration states: "If technology in this document is included in a standard adopted by IETF and any claims of any Huawei patents are necessary for practicing the standard, Huawei will not assert any patents against any party that implements the standard, however that Huawei retains the right to assert its patents against any party that asserts any rights against Huawei; and Huawei retains the right to assert its patents against any product or portion thereof that is not necessary for compliance with the standard."
News of the patent application has led to a virtual suspension of work on the standard, resulting in a long discussion of what to do.
First of all, it is distracting workgroup members from doing actual work. There is a lot of discussion on the patent issue, but much less on the actual work to do. It might even happen that the current approach syslog over TLS is dropped in favor of an alternate solution (syslog over SSH). This alternate solution is less used in practice and seems to be more complex to implement (the latter is my personal opinion). In the worst case, Huawei's move could cause syslog standardization to fail. This is because the working group is already very delayed in delivering its products. It needs to finish its basic tasks by the end of the year or it will most likely be terminated by the IESG (the decisionmaking body of the IETF). The patent claim definitely causes some extra delay.
No one seems sure exactly what Huawei is patenting. Gerhards praises David Harrington's contribution to the syslog standard but says he has recused himself from the discussion over what to do over the patent application. This includes stating what it is that Huawei has filed a patent application for.
Gerhards doesn't have any insight into the topic of what it is Huawei has actually filed for, either, saying:
Quite honestly: I do not know. Huawei's people have definitely done a good job in looking at what's currently deployed, what the syslog community wants to see, discuss some technical details (framing and certificates). The technical content of their paper is about four pages, so there isn't even much substance in it. Everything inside these four pages has been discussed on the IETF mailing list and the content almost exclusively stems back to other people's comments.
Frankly I am not bashing here I just can't find any novel contribution. Other WG members have similar problems. Huawei still says there is something novel, but they are not disclosing it.
Gerhards believes that Huawei wants to settle this issue with the IETF. "The license they are offering, as well as the way they proceed, indicates that they do not want to hinder the IETF process. Huawei employees contributing to the working group seem to have some influence on their employer to solve the situation," he says.
He goes on to say that the patent is being claimed where nothing new exists. He cites this Usenet post as an example of the technology being standardised in use as early as 1999, and this LinuxJournal article from 2001 on the same topic: syslog with SSL.
"The other problem," Gerhards goes on, "is that Huawei might change its license, or sell it (maybe as part of a merger), in which case every work based on the substanceless patent again is in danger. As such, I expect that the patent claim will at least stop open source developers from implementing the so encumbered standard, no matter how liberated the licensing terms may be."
Who is Huawei and why do they want to patent an unspecified part of a notyetcomplete syslog protocol?
Huawei is a roughly 34,000employee privately held telecommunications company based in China that is active in most markets outside of North America. One of its major competitors is Cisco, which once accused Huawei of stealing its intellectual property, settling the lawsuit in 2004.
As of press time, Huawei had not responded to a request for comments emailed to the contact address listed on the patent disclosure asking for Huawei's side of the story, and for comments on what, exactly, Huawei is patenting and why. We will be sure to pass on any responses we might get from Huawei.
Originally posted to Linux.com 2006-06-29; reposted here 2019-11-24.
words - whole entry and permanent link. Posted at 18:29 on
June 29, 2006
Tor: Freedom for whom?
Proponents of Tor recommend reading renowned security expert Bruce Schneier's article on the value of privacy. Schneier makes a compelling argument in favour of the value of privacy. But use of Tor isn't just about privacy.
There are, fundamentally, two forms of freedom. There is the freedom "to," and the freedom "from."
There is also the balance of freedoms: how one person's freedoms affect another's. Services like Tor address both the freedom "to" and the freedom "from," but deprive others of both freedom "to" and freedom "from."
Tor works by routing a user's Internet connection through a long and wholly undocumented and unlogged list of participating hosts. Theoretically, it is impossible to trace a connection back to its origin through this system. With the lack of logging, the only practical way is to monitor participating hosts and try and figure out who is doing what. The result is that anyone who uses Tor is anonymous to anyone whose services he is using. This provides the Tor user the freedom to privacy, and complete freedom from being identified.
This also takes away service providers' freedom to monitor access, and the freedom from abuse.
Bruce Schneier's argument, as twisted by Tor users, would appear to be that it is not a provider's right to know who is using its services. Tor users worry that providers are in a position of power, and power corrupts. The logic employed that if a provider knows who is using its services it will use that information for nefarious purposes is no more sensible than assuming that someone who is using a privacy service like Tor is necessarily doing so to facilitate troublemaking.
My fundamental problem with Tor is connected to my experience as an IRC operator. On IRC networks, Tor prevents freedom from abuse. If a hundred people use Tor, and one of them abuses his privileges on a provider's network, the only alternative for a provider (other than allowing the abuse to continue) is to block all 100 users, because there is no way to differentiate among them. Because blocking large groups of users often is not a practical solution, that one problematic user can continue being a problem without any limitations.
Privacy vs. freedom
Schneier states that the debate is wrongfully categorised as a debate between privacy and security. I agree it is not privacy versus security, it is privacy versus freedom. When one person's privacy restricts someone else's freedom, we have a problem.
In the real world, every country has a legal system with a set of rules by which everyone must live. If someone breaks one of those rules, a police force and judicial system exists to prevent them from continuing to do so. In some cases, the rules are unjust, but generally, rules are designed to protect the freedoms of others. Take the police force and judicial system out of the equation, and you end up with anarchy.
That's what Tor brings to the Internet. If everyone on the Internet used Tor, and no one could figure out where anyone was coming from anymore, the Internet would be a complete anarchy, even though most people would still attempt to continue their normal, honest behavior.
While IPaddressbased restrictions may not be an ideal solution for managing services on the Internet, it is the best currently available. Tor in effect removes this system from the Internet.
Prior to Tor, similar problems existed through open proxies and hacked accounts, but these can be blocked, because there is no such thing as a legitimate user coming through these means.
Please understand, I'm not against the concept of privacy. What I am against is the concept of total anonymity. I would not object to Tor, or any other anonymising service, if it provided a way of uniquely identifying users. I don't care if connections can be traced back to actual end users, just that they can be managed separately. But making end users identifiable is contrary to the stated objectives of Tor.
Are there practical solutions? Yes. The simplest solution would be to require registration of Tor users, and have service providers implement a system to check users' registration status. Though it wouldn't eliminate problems, it could reduce them and make them more manageable. Unfortunately, it would remove the very anonymity Tor seeks to create.
Is there a way to balance the privacy of users with the propensity for bad apples to destroy the crop? If so, what is it?
Originally posted to Linux.com 2006-06-24; reposted here 2019-11-24.
words - whole entry and permanent link. Posted at 18:26 on
June 24, 2006
April 25th, 2003 (from Advogato)
FIDO's GPRS service runs flawlessly in Linux. I'm thrilled. Here's how:
Use pppconfig or something, it's a standard dialup connection with numer *99# and l/p both 'fido' - authentication is really done using the smartcard in the GPRS pcmcia card. As far as Linux, well pcmcia-cs, is concerned, the GPRS card is a totally generic serial modem, so treat it like one.
I spent a while looking around for information about GPRS support in Linux and while I figured it existed I found nothing concretely useful. So maybe someone will find this useful.
I called fido up and asked for the ppp information to use my gprs card in linux. I was put on hold for a moment, and a technician came on and told me everything I needed to know.
By contrast, when I've had to call Rogers up for problems now and again with my cable modem connection, the moment I say Linux they literally hang up on me. Some service.
words - whole entry and permanent link. Posted at 23:45 on
April 25, 2003
November 30th, 2002 (from Advogato)
An entirely uneventful month has gone by.
For the seven years I've had a webpage somewhere or other, I've never had any qualms about revealing personal information on the Internet. Why not learn who it is you're talking to or about? Why lie about who you are or where you stand? So what if someone knows where I live? Anyone with a registered domain is already susceptible to name/address/phone number searches and if anyone cares badly enough, they'll just stalk you and learn everything about you anyway in a far less pleasant way than by reading your web page.
So I've updated my web page to discuss who I am, where I stand, and what I've done, though I've left some of the lame humour that has been there for years for, er, historical purposes.
words - whole entry and permanent link. Posted at 23:42 on
November 30, 2002
(RSS) Website generating code and content © 2001-2020 David Graham <firstname.lastname@example.org>, unless otherwise noted. All rights reserved. Comments are © their respective authors.