Dave Naffziger's Blog

We’ve been wrestling with a difficult decision for awhile at BrandVerity - how much information should we share about the techniques affiliates are using to hide from our customers?

Our company’s focus is on detecting affiliates that are violating their agreements with their merchants. These agreements typically disallow very specific activities (purchasing the merchant’s trademarked terms on paid search).

We certainly want to provide detailed insights to our customers but we also want to be careful of teaching would-be-abusers how to improve their techniques. This balancing act is a challenging one for us and up until now we’ve kept any discussion of the advanced techniques used mostly out of public discourse.

We decided to change that, and have posted our first detailed discussion of a technique used by affiliates to evade detection to BrandVerity’s blog. We also sent a more detailed version of this post to our customers before the post went live on our website.

Over time, we’ve seen the techniques used become more advanced and more widespread. We’ve seen deep and detailed discussions on blackhat forums, and even seen the launch of paid services that assist new abusers.

We haven’t seen a commensurate increase in industry awareness. The discussions on most whitehat forums continue to focus on the more basic techniques, and there is little to no discussion of them in educational resources available. Our intentions aren’t entirely selfless - we expect these posts to not only raise awareness of the techniques, but also to increase awareness of who we are and what we do. But that incentive has always been there and if anything is less impactful now than it might have been earlier in our growth. Ultimately, we felt that we have recently seen a shift in the balance of information too far in favor of the abuser.

Our intent is now to continue to discuss the tactics used in an open and transparent manner. This was not a decision we made lightly and we may reconsider it in the future.

In tech circles the belief that consumer ISP DNS servers are terrible is widely accepted. So, I found this study fascinating. The quick takeaway is that Content Distribution Networks (CDNs) like Akamai rely heavily on the location of the DNS server to select the server to serve content from. Websites that rely on CDNs to serve some content load slower because the CDN chooses a suboptimal server. And nearly every major website uses CDNs for some of their content.

I ran the test script from my home comcast cable connection. In my case, the CDN server returned by OpenDNS was the slowest in both tests, while Google’s was the fastest. Here is the data from my test of Akamai’s CDN (summarized for readability):

>>> test(‘profile.ak.fbcdn.net’, ‘Akamai’) == testing : profile.ak.fbcdn.net Akamai =======
testing : OpenDNS ( 208.67.222.222 )
204.2.171.82

Ping statistics for 204.2.171.82:
Approximate round trip times in milli-seconds:
Minimum = 46ms, Maximum = 48ms, Average = 47ms
—————————————————
testing : Google ( 8.8.8.8 )
96.17.69.66

Ping statistics for 96.17.69.66:
Approximate round trip times in milli-seconds:
Minimum = 11ms, Maximum = 13ms, Average = 12ms
—————————————————
testing : default ( 68.87.69.150 )
96.17.70.16

Ping statistics for 96.17.70.16:
Approximate round trip times in milli-seconds:
Minimum = 36ms, Maximum = 38ms, Average = 37ms
—————————————————
=================================

On closer inspection, the Akamai node that serves my content varies by DNS solution used (I’m in Seattle):

DNS Provider	Akamai Node Used
OpenDNS	Los Angeles
Google	Seattle
Comast	Seattle

Conclusion

The author of the original post makes a great point. The location that CDNs use to serve content impacts browsing experience much more so than the speed at which DNS resolves an unknown domain. The DNS server used for the request can have a big impact in the selection made.

In my specific instance, Google’s DNS sends me to the best CDN node (at least for Akamai and Internap)

An IP Geolocation Aside

At Quova, we primarily sold against Akamai’s EdgeSuite service which provided some geolocation data. One of the common myths that Akamai sales staff liked to propagate was that Akamai’s data must be better because of the thousands of Akamai CDN servers.

We would usually respond that CDNs primarily depended on network topology, not user geography. Therefore the many CDN servers Akamai used provided no benefit to the core IP geolocation data. I think if I had known how much Akamai depends on DNS sever location, our counter-arguments might have been stronger.

I’ve been using a few services recently that I’m pretty excited about and wanted to share. I wanted to give a bit stronger of a testimonial than I could in a tweet so I’ve been collecting them into a blog post:

Interesting

• The biggest cloud provider is a botnet. Google manages ~1M CPUs, while the Connficker worm controls ~18M CPUs. Holy crap - this gives me an entirely new perspective on botnets.
• Snake Oil Scale. A great graphical look at various supplements and the extent to which science backs up their efficacy.

Cool

• TechStars Seattle Launches! It is great to see the Seattle entrepreneurial community come together to launch TechStars. The team behind this is fantastic. It will become a great launchpad for Seattle startups. Also, applications are now open.
• Seattle 2.0 New Management. Marcelo has done a fantastic job building Seattle 2.0 into the nexus of all things startup in Seattle, and I’m excited to see what the future holds for the brand!

Useful

• Rapportive: The best gmail firefox extension ever. Pulls your social media info for your email contacts into the sidebar (replaces the ads). It keeps creating serendipitous information discovery.
• HgInit: Stunningly written documentation on Mercurial by Joel Spolsky. I found the transition from svn to mercurial similarly difficult to Joel’s. This guide helped immensely. Many open source software suffers from terrible documentation. I wonder what other open source movements could achieve with similar efforts.
• Athlinks: Automatically gathers all of your race results (ok, most of them). On sign up, you are presented with a series of events and you check the ones that are yours. I’ve actively wanted this built for a long time. Not perfect yet, but I love the vision. A great showcase for what ‘intelligent agents’ (and likely a little bit of manual config) can do.

Marin Software and SMX ran their second annual ‘Biggest Search Geek’ contest a few weeks prior to SMX West.

While, I barely run AdWords campaigns (just BrandVerity’s in-house campaign which is frankly way under-optimized), I was mostly curious about what types of questions they asked and how much I knew (I was already speaking at SMX West so the prize had no value to me).

As it turns out, I must know more than I thought. I got an email over the week from Marin Software that out of 1200 testers, I placed in the top 50.

I even get my own linkbait badge =).

Every year my wife sends photo cards to our family and friends for the holidays. She carefully sorts through our photos from the year and selects several that she shares on the cards. She shuns the cards on photographic paper and instead selects heavy stock folded cards. The final product is usually great, however we’ve definitely been disappointed by print quality from time to time.

This year she used Tiny Prints (after a great experience with their birth announcements). Everything about the card was just better than the ones we had ordered in the past. But, the single biggest, unexpected awesomeness was the photo touch up they did on the cover photo of the card.

See the before and after below (yes, one version is scanned from the card). Sophie had a slight, but noticeable pimple to the right of her mouth in the original photo. It wasn’t something that we had thought to touch up, but we absolutely noticed that it had been removed when we got the actual card.

The amazing thing is that this is what Tiny Prints does. See co-founder Ed Han’s description from their ‘Our Story’ page:

Since our first year together, we’ve reviewed every order, edited every photo and obsessed over every typo and etiquette mistake on every card that has passed through our hands. After years of extreme perfectionism, our customers expect this level of attentive detail, and every member of our staff is proud to provide it.

I don’t think I could have been convinced in 2003 that there was a market for another online photo card company. Ofoto, Shutterfly and Snapfish seemed to have that market pretty locked up. However, there was clearly a high end to the market that had been overlooked and Tiny Prints has grown aggressively since. They initially launched with a focus on birth announcements, and have steadily expanded to a whole range of cards, calendars, photo books and even corporate cards (no save the date magnets though).

At this point, I can’t imagine we’d order cards from anywhere else. There aren’t too many businesses that fall into that category.

Startups rarely launch the way they are intended. Jeff and Todd had spent the better part of a year working on the groundwork for their apparel company, Out of Print Clothing.

They had spent months obtaining license rights for the original cover art from a number of cult classics such as Atlas Shrugged, Brave New World and On the Road. Tracking down the artists that had created artwork 80 years ago took a bit of work. In one instance they traveled to an old mansion in Brooklyn to meet with the widow of one of the artists.

Both Jeff and Todd were pursuing a certain look in their shirts. The quality of the shirt and silk screening was essential - they tried lots of shirt/printing variants until they were satisfied.

More importantly they struck a relationship with Books for Africa to donate one book for every shirt purchased.

They had a launch date planned for several weeks down the road. Everything was coming together nicely. They were lining up influencers to wear their shirts and had developed a media list to reach out to.

Then JD Salinger died.

And they had license rights to the original cover art from Catcher in the Rye:

They had a few more weeks of solid work before launch, but knew they couldn’t miss the opportunity. They worked through the night, polishing as much as they could, cutting requirements where they could and launched the day after the news broke about Salinger.

The result is pretty awesome. Check out their line-up of shirts. In addition to the JD Salinger shirt, these two are probably among my favorites:

Oh and of course, they have the cover art to this post’s namesake:

Mint.com ran an interesting post on selecting a 529 plan written by Matthew Amster-Burton a finance-food writer based in WA State. Without surfacing much research, he makes the quick assessment that the Montana 529 CollegeSure CD is a better low-risk investment than the WA GET Plan (and he has even invested in the GET Plan).

I wasn’t too familiar with the Montana CollegeSure CD, so I dug into it a bit. The specifics are:

• For new investors, it returns a minimum return of 2% a year
• Returns a maximum return equivalent to the growth of tuition at the IC500 less 3%.
  • The IC500 is a basket of 500 independent colleges.
  • This rate was 4.28% from 09-10.  It looks like it was roughly 5.5% over the last 5 years
  • Over the life of the index (1983-present), it has grown by an annual rate of 5.9%
• Can be purchased with maturities up to 22 years
• FDIC Insured (the current limit is $250K)

So, over the last 5 years, the plan has returned 2-3% a year to investors.

WA GET bid-ask spread aside, the index of the WA plan (UW Tuition) has been growing at 7% a year. This is not uncommon, College Board data shows college tuition growing faster at public institutions than at private colleges.

The FDIC is a more credit-worthy institution than the state of Washington, and the Montana plan isn’t subject to legislative risk in the way that the GET Plan is.

I personally assess the likelihood of both the plan and state of Washington defaulting to be very low (not as low as the risk of the FDIC defaulting), but still low enough that I consider it a wash. I don’t know how to evaluate legislative risk though - that is something endemic to the WA Plan.

Well, sort of. One of my articles in FeedFront was republished in Internet Marketing from the Real Experts. I guess I can now call myself one of the ‘Gang of 88’. I’ll have to give more thoughts later on my newly acquired gang membership…

Although the distribution of the book will likely be similar to the magazine, I can’t help but feel that getting published in a book is more substantial. I guess old media habits die hard.