I have a ClusterHat on a pi3. I am rather looking forward to using it on a p4 since memory space was primarily the issue I had with the controller. The 4 zeros are fine because I don't run a gui on them. The extra speed between the bunch of them will probably go a long way, too.
I'm on my 9th class, one more to go. I found that Machine Learning had decent involvement from Isbell (he'd chat with us on Slack) and that Operating Systems and HPC and excellent professor participation -- they would hold weekly office hours (video conferenced) and were active on the forums.
There were 1 or 2 of my earlier classes that did feel phoned in, though (but did have strong TA involvement at least).
Actually it's much better than hipmunk. I used to be a fanatic user of hipmunk. But once I tried google flights i immediately was amazed by its speed, browsing different dates is super fast and easy.
Google Flights' instantaneous search allows you to freely explore the option space opening up new ways of travel planning. Free for spring break and don't know where to go? Set your travel dates and pan around the map looking for deals.
Kayak doesn't even come close to this. At best, you can set the month of travel; useless for most busy people.
As with all things Google, it's an order of magnitude beyond what hipmunk does. If you keep the window open long enough you will actually see flights disappear if they fill up.
I still think the hipmunk interface is more humane.
There's presenting the existing data well (Google) and there is adding value to existing to make it more useful (Hipmunk).
I find that Google's innovation for almost every one of their products is in saying "we have oodles of data, how can we display it effectively"? They have done well at collecting and quickly accessing large amounts of data, linking loosely coupled sets, and in some cases, adding meaning (image searches understand the concept of "bright" or "dim", for example).
But I rarely see them taking the data they have and merging it in a way to create new meaning or value. Hipmunk, beyond co-opting a gantt chart, did this with their "agony" score. Google not only didn't try here, they didn't try for local search either, and just acquired Zagat for their score (and brand name and loyal following, etc.). And many startups offer meaning above and beyond that combined in the data: through their experience, their ability to merge context from being subject matter experts, or just their flexibility to experiment without a spotlight on them. Google tends to stop short of any of these.
So, yes, it's no Hipmunk: it's a typically well done Google product taking hairy data, merging it up, and displaying it well. Doing more than that, however, is left to the rest of the world.
(Though, one wonders why ITA didn't do some of these things before when they first created this better way of looking at flights a few years ago...)
Both Hipmunk and Google Flight Search help you sort out bad flights in ways that other websites don't - Hipmunk calls this "agony". To see this in Google Flight Search, scroll down to the bottom of a results list, you'll see something like "123 longer or more expensive flights hidden. Show all." - these are the longer, no less expensive layovers that you probably don't care about.
Longer or more expensive is simply filtering. Agony is creating a new metric. I see those as different.
Now, if the metric is useless or synonymous with some of it's underlying drivers, then you are right, nothing there. I find the agony score to be more useful than any small combo of filters, but perhaps that's just me.
ITA licensed their engine to the whole industry, instead of competing against it. So they focused on what they did well and let Orbitz et all work on UI and customer service.
Many companies start there, and then start competing against their licensees. Google is a great search engine for finding credit card deals, and they now recognize that and offer their own credit card aggregation engine: https://www.google.com/advisor/uscredit Looks a lot like all of the other aggregator sites (creditcard.com, etc.) out there, right? Now, compare to CreditKarma (http://www.creditkarma.com/) which attempts to change the process. Again, Google does great work, but I think they stop short at times when our industry really needs to be pushed. Startups can only do so much.
BTW, ITA released their "matrix" front end to the world (http://matrix.itasoftware.com/) around a year or so after Orbitz and others licensed it. The clicks still took you to externals to complete the purchases, but new filters and capabilities were often tested on the live ITA site before 3rd parties took advantage of them. The problem I always felt was that they were just new filters: fast, live-response, rapid return filters on very large and nasty data, to be fair, but still just filters. I enjoyed sites that added additional context to the flight experience, few though they have been.
Again, thanks to both Google for organizing hairy data, and startups for doing more with the data. Glad both have room to play.
I love hipmunk but they are resting on their laurels a little too long if you ask me, you still can't edit an individual search tab (I asked and was told to fiddle with the url values to edit the search). And you still can't move tabs around, plus Google flight actually makes it look really slow.
That being said, it is rare that I can find a better rate than what I see on Hipmunk, and their agony ui is still frickin' amazing.
I tried using Hipmunk a few times, but the prices were always at least a hundred dollars more than if I had just gone to Orbitz.com and dealt with their crap interface--I've always found that Orbitz has prices at least as good as anywhere else, so I tend to judge against them.
Google, on the other hand, gave me exactly what I needed at prices competitive to Orbitz. It was a bit disappointing that when I was done, I had to buy the tickets through the carrier's site, but it still found me a flight at a good time for a good price.
I like Hipmunk's interface better than Kayak's, but Kayak seems to have better data: there are sometimes good fares that appear in Kayak but not Hipmunk, while I haven't found the reverse to be true yet.
We get our data from mostly the same sources, so this should almost never happen. (There are also lots of times we have better rates than Kayak, e.g. on Amtrak and Airbnb, which they don't have.) Next time you see this, could you email contact@hipmunk.com?
Love that you put in date ranges in Hipmunk, since I am usually not sure which days have the best deals and am not strongly tied to returning/leaving on a certain day
"Art for Art’s sake is an empty phrase. Art for the sake of code, art for the sake of the terminal and the OS, that is the faith I am searching for." - Salvador Dali
