http://aws.amazon.com/ec2/#pricing

After a year of free you can buy a reserved micro instance for $23/year and pay about $8.76/month (for a total of $10.68/month for a year).

daveT,
Thanks for the post man. I kept noticing that thread and I've never bothered going through it, but it does look helpful. I'm gonna dive into it tonight.

I am happy to report that I've done a little better than you predicted. I was making a couple of mistakes that I'll explain in case it could be helpful to someone else.

I won't link to my site again because I feel like I'm spamming if I do*, but it's a somewhat specific niche -- web-based retail loss prevention management software. I began by just adding tons of keywords, all of which seemed to be good for me. Terrible idea.

In a nutshell, using tons of keywords, many rather broad, was completely worthless. For example, "loss prevention" seemed like an obvious choice early on. It resulted in like 1500 impressions, zero clicks.

What does seem worthwhile is using very specific keywords that are narrowly tailored to your offering. The second component to this is to write ads specifically in response to those keywords, but not in a spammy sort of way. This seems to have a few advantages:

1. Greater rarity implies likely less competition, resulting in greater likelihood of favorable page position and fewer ads vying for attention.

2. Rare keywords suggest fewer good search results, possibly increasing the likelihood of a user resulting to clicking an ad.

3. A specific, narrowly defined topic seems to create a situation where a well written ad, tailored specifically to that user in that situation, may result in a more valuable click from someone seeking your offering. The non-spammy appearance seems crucial for this to occur.

I have some evidence to suggest this works -- a very specific, highly relevant phrase for me is now resulting in a 15% CTR (of course sample size, etc.).

I'll keep everyone updated. Anyone else have any experience/input?

*Need to save my spam goodwill b/c I was planning to ask you guys to critique/review my site and app both.

I should add that the phrase I was referencing is 4 words long. It seems that 3 words or 4 words is optimal for the strategy I outlined.

in addition, another advantage to narrowly targeted adgroups to a single phrase - the search text is bolded in your ad leading to usually higher CTR.

15% CTR is more like it!

That's the way to do it - bulk test a ton of keyword phrase + ad combos, keep running those that hit high CTR = profitable search campaign.

Remember the Google uses your CTR as the main factor in what they make you pay - if you have a bad keyword in a group that just wastes impressions, it can easily drag into unprofitability otherwise profitable keywords. Conversely, with an awesome ad on a tightly targeted keyword with great CTR history, you can end up in position #1 of page 1 paying 0.05c/click where suggested bid is >$1

I didn't even think about that. That's probably one of the strongest reasons to take that approach. I'm such a noob at Adwords, I knew that from reading it a while back, but I forgot all about it until you just said it.

And i def know that 15% is just variance that will certainly flatten out. But I have another keyword that's at 9.5% CTR, which leads me to believe that even after they do revert back to the mean, it'll still be at a decently high (but more reasonable) CTR.

And this little guy is killing it:



I picked up the "for experts" line on a copy writing blog once. Apparently it works by creating a sense of exclusivity that people want to belong to. Evidently it works on anything.

"my ads are filled with lies i stole from other ads. why won't anyone click on them?!?!"

Hahaha wp, I laughed pretty hard at that.

Sort of like how my company decided to place "Lowest Prices Widest Variety" on their ads when neither assertion is true.

"Oh, but it has a ring to it."

I fought really hard to kill that one.

I was only laughing because my wasn't a lie

Now that old.sharkscope is completely broke, I find myself in need of drawing nice plain looking graphs from a series of results e.g. 1,2,3,4,3,4,3,4,5,6,7,6,7 ... - Can anyone recommend anything easy to use (windows dll / command line .exe / php function) that I can use to make automated "old sharkscope" looking graphs output as a PNG? gnuplot seems possible but rather complicated!

Just got back from a holiday at a ranch in Canada. Was a 6.5 hour drive north of Vancouver and I didnt even make it half way up British Columbia. This country is just way too big!

Anyone getting the new iPhone 5?

I recall reading somewhere that this is why so many 2+2 titles end with "For Advanced Players"

Well once you start to change stuff and need to see if other things break or once components get combined I'd rather have the test

Depends on your definition of math. "Real math"(tm) is pretty much like programming or art. You work on pretty hard problems, grind your teeth...maybe iterate...if all else fails take a walk...do something else and after 3 years you solve the stuff. Ok maybe more like long term programming or something

I forgot what it was called but that dude in Texas had it right imo....just give students stuff to work on and let them think about it. Explain principles let them work from there...no handholding textbook style.
If I could change anything about my education I'd most certainly want to give this a shot...sound pretty awesome.

I'm on Virgin Mobile, which just introduced 4G and the 4S. I'm loving my $200 'Droid tbh.

What is this lawsuit all about now?

http://thenextweb.com/apple/2012/09/...ing-name-a-pl/

I'm trying to see how an apple from a fresh food joint is infringement.

I don't know either way right now, but I believe that only using wishful thinking is really bad. Why would I want to clutter up my code with danglers when it does nothing to help me achieve the end result? I think that focusing on the separation of the code base is a better approach, ie, you build each section as a stand-alone proto-primitive that you can later combine.

I dare say it: too much wishful thinking can lead to premature optimization.

The wishful thinking is really one little concept they offer up at the beginning of the course. Yeah, the ideas build on a notion of wishful thinking, but focusing on the short-comings of wishful thinking is sort of like using assert statements to describe the entirety of a TDD book, then calling into question all of TDD because assert statements aren't really effective.

The math I'm talking about is taking a few derivatives and adding vectors. Some of the stuff is cool at first, but instead of "Wow, cool, look at this new way of looking at a f'(x) vs d/dx!" I'm thinking more like, "Oh, wtf. Why do I care? I take the derivative and do some chain rule and that gets me an answer. Big woop, I can do momentum. Next!"

One thing I always hated was Trig. Just way too much memorization, which I am awful at.

I went with the Galaxy s3 for the bigger screen instead of a longer screen and because I'm not a fan of dumb software patents.

How stupid is this?

PostgreSQL 9.2 comes out. I dump my current 9.1 db, uninstall, and now I have 9.2 on the computer. I'm going to create the tables and install the triggers, try to connect, and report back.

But 9.2 is 4x faster! and is support JSON (?)! and...

clown,
I wouldn't toss the idea of having tests. I like having tests around because it's 1 less thing to think about. If you run the test and it passes then you know it works.

Also have you guys used this before?
http://www.khanacademy.org/math/algebra

Seems like a pretty solid site for learning whatever.

I've read good things about khan academy. I watched some stuff on linear regression and gradient descent on the site one time. It had some nifty graphics and I imagine it would have been easy to understand if I had any clue what he was talking about.

Supposedly they are backed by the Bill Gates foundation and Google. I just watched their TED talk earlier. Seems legit. I'm going to also use the site you linked before dave to hunt and peck topics when I need them. It looked good, thanks.

I have about 3 weeks to learn 3 years of HS algebra.

Yeah but I don't think I can make a preorder work sadly. I want Verizon since AT&T LTE is nonexistent in western Loudoun County. So I want to switch.

Willing to give up my unlimited plan because AT&T has gone to war with it anyway, and it's only really of use at this point if I ever wanted to force an early exit at some point.