The tourist map of flash drives (things.gnod.com)

112 points by no_gravity ↗ HN
I made this map of the 200 most popular flash drives on Amazon today. It looks a bit like a mandelbrot set :)

I made this for myself, because I wanted to have an overview of whats available. Do you think this is something people want? It could easily be extended to ssd-drives, monitors and whatnot.

79 comments

[ 3.4 ms ] story [ 158 ms ] thread
I made this map of the 200 most popular flash drives on Amazon today. It looks a bit like a mandelbrot set :)

I made this for myself, because I wanted to have an overview of whats available. Do you think this is something people want? It could easily be extended to ssd-drives, monitors and whatnot.

it doesn't work with IE 11

how would things look if you plotted just capacity on the y axis instead of capacity/price?

what algorithm do you use to make the points form a grid in the dense areas?

Yes, I don't have windows, so debugging IE is a bit painful. If there is demand, I would do it of course.

You can add #y=gb to the url, it will change capacity/price to price.

I just brute force the positions. In the first iteration, I go through all empty cells and choose the best. For all items. Then I search for items that can be swapped.

I'd be a +1 for IE 11 (disclaimer, I work at Microsoft on the IE developer tools but use IE at home). Regarding testing in IE, have you seen the http://www.modern.ie website? There are Windows VMs there you can download for free for any host OS that have all the different versions of IE.
(comment deleted)
How is the 1 tb flash drive on the far right lower than the 32 gb drives on the left?
Y-Axis is Gigabyte per Dollar
The axes[1] are really poorly named.

[1] English plurals can be funny.

That could be. Im from Germany. What would be good names in english?
See comments below.

(A cool thing btw. Just to make sure you don't get me wrong)

The one at the top is (according to the reviews on amazon) a counterfeit PNY drive that is only ~64 GB.

Hey no_gravity: this comment is "dead" and I don't know why https://news.ycombinator.com/item?id=7465983

Yeah, I saw it had some bad reviews. Not sure if i should remove that one. I wanted to see "just the facts" so i included all popular drives.

Strange, that the comment is dead. It is the explanation of this post, so what's wrong with it?

No idea, appeal to info(at)ycombinator.com
This is really cool!

Interested to hear how you got the price/size data out of Amazon? :)

They have an API called "product advertising api".
I think this would be more useful if it were a graph of capacity against price, rather than a graph of capacity per price against price - which is a bit weird - by default.
Yes, this is much more useful as it doesn't require multiplying every coordinate in your head.

It's still kind of broken though. How about actually plotting the drives at their correct position rather than stacking them neatly?

Capacity vs. Price seems perfectly reasonable. The PNY 128GB drive which is at the top of the original linked graph is nestled between two other drives -- one smaller and pricier per dollar, the other the same size and pricier per dollar.

"How about actually plotting the drives at their correct position rather than stacking them neatly?"

It would be one big mess of overlapping stuff. I can code an option for that if there is demand for it.

Yes it would have that problem, but it would have the advantage of conveying useful information. How about plotting smaller points and providing the image and info on rollover -- the tiny icons don't serve much purpose.

Differentiating USB3 vs. USB2 would probably be informative too.

Whilst such a graph would certainly have the advantage of displaying the data accurately, do bear in mind that it would have the disadvantage of not looking like a partridge.
YES, that would help as currently you can see 256GB drives beneath 64GB ones in the capacity axis (which is terrible).
This is way more useful.
it would help just to have the y axis labelled correctly. afaict the y axis label is the same for both graphs (the default and the alternate link provided as sibling answer)!
I think this would be more useful if it were a graph with capacity per dollar on the Y-axis, and capacity on the X-axis. When i am buying a USB stick, i am not thinking "I have 20 dollars to spend, what is the best way to spend them?", i am thinking "i need 32 GB, what is the cheapest way to get them?".
This would be better if the y axis were actually capacity and not $/GB.
Interesting. This seems to be one of the top requests. Im not quite sure how that would look like. Because there are only so many different capacities (1/2/4/8...GB).

Here is a try to make the y-axis strictly capacity based:

http://things.gnod.com/flashdrives/#y=gb

Do you think this is better?

(comment deleted)
Unfortunately, that still shows some 64Gb drives higher than 128Gb ones.
Yes, thats a different issue.

All I changed in the y=gb chart is the meaning of the y-axis.

You cannot perfectly sort things in a grid.

Not being able to rely on the graph being properly sorted makes it a bit useless...
Yeah, that's more useful. Making them not overlap causes some strange positions, but this organization makes it easy enough to find different tradeoffs. If I want a very large drive, I recognize that it will not be as price efficient, and a look to the upper right. If I want a price efficient drive, I follow the left contour until I find the size I want. If I have too much money, I follow the right contour.
No. It looks better initially, but it misses the point.

When you move your mouse normal to the axis marked "capacity", all your data points should have the same... capacity. So they should all be 32G, 128G, whatever.

If you decide you want to spread them out, sure. But I shouldn't move my mouse down from a 128G drive, and find I'm on a 32G drive.

have hte y-axis be log2 of the capacity?
The X and Y axis seem to be labelled the wrong way around for me.
It would be a more useful graphic if the axis were actually capacity and cost (as labeled). As it is now the Y-axis is showing us the capacity/cost. When looking for drive I'm usually not trying to maximize capacity for a certain budget (say by buying several 8G drives). More often I want a drive that doesn't cost too much that has a certain capacity.
Have you considered open-sourcing this? It's very cool but I think it could be improved.

I think a more useful view would be to show the flash drives in terms of price per GB, within a certain range. By plotting them in this discrete (best word I could think of) where they cannot occupy the same space in the graph it's misleading about the best value for money.

Not sure what to do with it by now. I'm really excited to see it reaching the front page. Maybe this is an indication that a "scatterplot-shopping-site" could actually be a good idea. Yes, there is a lot of ways this could be improved. Would be quite easy to add filtering and different views.
Cool, but (there's always a but) there's one parameter sorely missing -- speed. What's the point of having a humongous pendrive if you need two days to copy its contents?

Unfortunately it's not as simple as price and size, with the advertised speeds tending to be higher than measured. Also, a while ago I got a drive which promised pretty nice nominal read and write speeds; it turned out however that it had a small buffer and for larger files (or more of them) the transfer speed slowed to a crawl within a few seconds.

Tricky items, those little drives.

Exactly what I was thinking. I have a ton of pictures of the kids I wanted to transfer to my in-laws. My first inclination was to throw them on a couple of flash drives like these. In the end I just grabbed a 500 GB USB HDD as it was less of a hassle.

I also learned that while the cloud solves lots of problems for me, sometimes a flash drive is still the easiest thing to use.

Edit: Then again, sometimes speed doesn't matter. I just got a USB drive in the form of a fictional character that sits next to my other trinkets above my desk. I use it to back up my GPG and SSH keys in case my laptop goes kablooey.

> In the end I just grabbed a 500 GB USB HDD as it was less of a hassle.

Sneakernet wins again!

(comment deleted)
Performance/Bandwidth vs price would be interesting as well. But it probably correlates with $/GB.
This is incredibly cool!

Did you create the chart yourself or is it built on top of some kind of js library?

No library, just some hours of hacking away in plain javascript.
That's awesome, the result looks great :)
Bought a new drive last week (ended up buying a 64 GB Mach Extreme stick). The most difficult aspect is finding a stick that's fast.

The actual as opposed to advertised speeds vary wildly. Anything below 20 MB/s isn't really state of the art anymore, and the one I bought gets 60 MB/s. Drives with 100 MB/s are available; after that comes the very-fast-very-expensive USB3 SSD market. If you buy a drive without paying attention to benchmarks, you can easily wind up with a 5 MB/s dud.

For what it's worth: I couldn't find a micro usb drive (ie. about as small as the USB port itself) that had great transfer rates. The best I could find was -- I think -- a Corsair stick offering 20 MB/s. If you want 50 MB/s+, like I did, you're stuck with "regular sized" drives, for the moment; all the faster drives I found were even more bulky.

(comment deleted)
Is there anyway to add in the rating on amazon for weight? A lot of the drives that sit higher in gb per dollar have pretty terrible user ratings. If not weighted on the chart, at least the # of stars/rating in the hover information.

Then, if not ranked, in the information the speed to which the drive works?

Over all, really cool site! I love it.

The data comes from the Amazon api, which neither provides ratings nor speed. As for ratings, im not sure if it would be ok to scrape them. Probably not.

Speed is a really complicated issue, because afaik the speed can even be different between 2 drives that are sold as the same. The brands do not produce the drives themselfes, but brand drives from different manufacturers. So you can order 2 times the same drive and get 2 completely different ones. Except for price and capacity.

It seems like the axes are wrong.
Love it, but it seems like this would be a perfect case for using D3 rather than hacking away in pure JS, and a good excuse to learn D3
How would it benefit from D3?
I think the D3 suggestion would be it would allow the chart to be dynamic, changing an axis from price/gb <-> price
Very cool. The 1 TB drive off to the far right though, shouldn't it be way off the top of the chart? Assuming you made some kind of adjustment here to keep the outliers within a reasonable range though it looks like it's a 64GB drive.

Edit: Looks like the axis are inverted?

y-axis is gigabyte per dollar. if you want only gigabyte, it would look like this:

http://things.gnod.com/flashdrives/#y=gb

I have to make it more clear or switch to a different default view. Not sure yet.

Switching the label on the default from More Capacity to More Capacity Per Dollar would clear it right up and should be changed even if the default view isn't going to be changed.
I'm finding things to the left of things that are cheaper, eg Integral Neon 32Gb (29.18) left of Centon Datastick Pro 64 (29.97)?
Duh, typed the wrong example there. The Leef Supra (24.99) is to the right of the Integral Neon, but less expensive.
Yes, you cannot sort them "perfectly" without overlapping.
I would be really interested in the amount of amazon affiliate money a site like this can create. If you are willing to share (perfect opportunity to boost it to the frontpage again in a month) I would be grateful!
Sure, can do. Im taking guesses :) Since it only links to amazon.com im not sure it will generate too much. Amazon doesn't have an international affiliate program. You have to sign up for every country individually. If this rakes in tons of money, I will do that of course :)

Also some punishment seems to have kicked in. This post suddenly jumped down 20 positions. Posts older and with less points are above it now.

Mouseover is a little screwed up for touch devices. Nice job though, very useful.
I now added a handler so the tooltip disappers when you click or touch anywhere outside of it. Does this solve it?
Thanks! I think this could use a date as to when the data was collected, or if it is always up to date I suggest you make a clear statement to that effect.
Yes. Interestingly, I just found your comment downvoted on the bottom of the page. I wonder why, because I think your point is really important. Im experimenting with the speed of updates and will have it explained on the page as soon as I have it settled.