Monthly Archives: October 2012

iBooks 3.0 fixes two major gripes — mostly

Shout it from the rooftops, iBooks 3.0 is a gift from the Apple gods.

I use, in this order, iBooks, the Kindle app, Files Pro (for PDFs, but iBooks may be taking on that role), and Stanza (not so much anymore). I prefer iBooks over Kindle for a few reasons: 

  • The store integration is much better. In the Kindle app it’s non-existent, instead you have to go to Amazon’s web site. This has gotten better, but for some time it was far too easy to find the print edition of a book and not be able to find a way to purchase the Kindle edition. Worse (at least the last time I bought something) your purchases went into a funky folder in the Kindle app, and every. single. time. I would forget where it was, and have to hunt around to find the thing I’d just bought.
  • The mechanics of flipping pages is cleaner, and in particular I am able to make it so tapping either side of the page goes to the next page (going back requires a swipe, but that’s the much less common use case).
  • The overall look is nicer. One specific example of this is the highlighting: iBooks’s highlighting is more variable, and looks more organic. Kindle’s looks more programmatic.

But iBooks had two glaring, annoying, really-frustrating flaws:

  • Highlighting across page boundaries was nearly impossible. There were two ways to do it, one of them unreliable, the other a slow pain in the ass. 
    • The unreliable way is to drag the end of the selection to the bottom right of the page and then hunt around. At a certain spot, the highlight extends a little bit, as if to indicate that it has gone on to the next page without you (which it has). This is neither exact, nor consistent. Sometimes it highlights just the first line of the next page, which is preferred, albeit inefficient, because you then have to go to the next page and open the highlight for manipulation and extend it. Sometimes, though, it highlights the whole next page, and this is not undo-able. The only solution at that point is to start over.
    • The pain in the ass way is to highlight up to the end of the page, then redefine the font size. This changes the page breaks, possibly making it so you can highlight more text. This is a pain in the ass because each time you change the font size, the entire book re-paginates, which takes (on my iPhone 4S) anywhere up to about twenty seconds, and then has to be re-done of course when you switch back. You can ignore that it is happening, but until it’s done navigation is a bit wonky.
  • There was no way to copy even a small portion of a book to send to a friend, share, or just to note separately from iBooks.

Fortunately, both of these issues are (mostly) history as of iBooks 3.0.

With the new continuous scrolling in iBooks 3.0, you can easily highlight whatever range of text you want. Unfortunately, it appears that performance (again, on an iPhone 4S) is abysmal when in continuous scrolling mode. I mean unusable. I mean Apple Maps bad — swipe the screen to move the text, and wait for five seconds before *anything* happens. This seems to go away after a few minutes, perhaps once iBooks has done whatever it needs to do to prepare for you scrolling through the book, but until then you should just read the same paragraph over and over, because you’re not going to the next screen.

NOTE: If you don’t like continuous scrolling, then tough: the situation has gotten even worse for you. It appears that the unreliable method listed above for extending a highlight from one page to the next no longer works. That leaves the “change the font size and hope for the best” method listed above.

Now, when you select text in a book, you have the option to send through any of the usual suspects, including Message or Mail, or you can simply copy it, and do whatever you like with it. The selection is quoted and cited, with a link back to the iBooks store for the book in question. I don’t begrudge this at all; you can delete it if you like, and it’s a useful feature to give whomever you send it to a link to get the full source.

In summary

If you like continuous scrolling, this update is full of awesome. Highlights finally work in a reasonable way, and sharing/copying is now possible. If you prefer pagination, then this is a two-steps-forward, one-step-back situation. The ability to copy is oh so nice, but the highlighting in paginated texts appears to be even worse than before.

Advertisements

Why Multi-armed Bandit algorithms are superior to A/B testing

This is a really nice description of the benefits of Multi-armed Bandit algorithms for testing: chrisstucchio.com

In summary: multi-armed bandit algorithms allow you to automatically present to your users whatever is the most likely winner among your options on every single test.

The huge advantage is lowered regret: you’re always showing your best guess at the winner, so you show the losers less often.

In addition you can overload your testing queue as much as you like, with the only consequence being that the algorithm will be correspondingly less certain that it is doing the right thing. But it will always make its best guess.

The drawbacks include that it’s not inherently built to consider traffic variations due to daypart, seasonality, different promotions, etc. You can reset the statistics anytime you like, but as it is presented here the algorithm isn’t set up to compare Tuesday data to other Tuesday data, for example.

Even so, this seems like a huge step forward from traditional AB testing for any outfit that is aggressively optimizing. I remember talking to a multivariate vendor who touted his system’s ability for you to pick only the combinations you wanted to test, and then to terminate early losers. That’s an inferior attempt to accomplish the same things multi-armed bandit achieves.

Negative ROIs are Ridiculous

Suppose you have a printing business and you spend $1000 on advertising in January, and you make a gross profit of $1010. Per wikipedia’s ROI article, ROI is:

Return on investment (%) = Net profit ($) / Investment ($) × 100

or: ROI = (1010 – 1000) / 1000

That’s 1% for the month, or annualized about 12%. Not bad — it means that you’re doing better by buying the advertizing than putting the money in the bank.

So far, so good, but what if you’re not profitable yet? Worse, suppose you are incredibly un-profitable. Suppose you spend $1000 on advertising and make only $10 back. By the same formula you get:

ROI = (10 – 1000) / 1000 = -99%

That’s -99%. If you understand ROIs that’s fairly straightforward, but here’s the rub: suppose that in February, undeterred by your lack of success in January, you spend another $1000 on advertising. You’re still not very successful, but you do manage to make two sales and a gross profit of $20. Back to the formula:

ROI = (20 – 1000) / 1000 = -98%

Take a moment to consider: your sales doubled. That’s huge. But the ROI went from -99% to -98%. That’s disappointing. The difference appears small becuase ROI isn’t measuring what intuiitively we think we measuring. ROI isn’t measuring the distance from 0; it’s measuring the distance from complete failure — from -100%.

Looking at it that way brings the example into line: you doubled your sales, and (-98 – -100) = 2 is twice as great as (-99 – -100) = 1.

The situation is just as strange around breakeven. How is ROI useful when comparing a month where you made a gross profit of $900, for an ROI of -10%, and a month where you had $1000 in sales, and hence an ROI of 0%. How does 0% compare to -10%. It’s better, obviously, but how much?

ROI is useful when you get a reasonable return and want to compare it to other ways to invest your money. If they ROI is better than 10% annually (roughly) then you should definitely invest.

A better alternative is to simply compare the net profit to the investment. It’s a small distinction, and perhaps it has flaws of its own, but it would show much better in cases like these examlpes what’s going on.

 

I’m curious what others would use in this situation

This is frustratingly misleading: Global crisis ‘has destroyed 40pc of world wealth’

WEF 2009: Global crisis ‘has destroyed 40pc of world wealth’

This is so frustrating. (note, the quote is from 2009, but I just came across a present-day article referencing this) The world’s wealth is our infrastructure, our skills, our goods and our labor. 40% of that did *not* go away. We did not blow up 40% of the world’s buildings, we did not salt 40% of our fields, we did not trash 40% of our clothing, cars, TVs and computers, and we did not suddenly vanish 40% of our people.

Putting it another way: the Fed could print a Trillion dollars today and there would still be children in America going to bed hungry tonight.

40% of our *money* went away. How exactly that happens is a little puzzling as well, but it’s a measure of the faultiness of money as a metaphor for everything else that something like this can even happen, and a measure of the danger of the metaphor that supposed experts can say something like this with a straight face.

Actually, I should qualify that. The last time I attacked a stupid quote like this, I happened to be able to trade emails with the expert in question. He was very gracious in the face of my rough-edged criticism, and explained to me that his actual paper had said none of the things I was criticizing him for. It was the news writer’s lack of understanding showing through, not the researcher’s (supposed) ineptitude.

So I’ll simply say that *if* this article accurately represents what these economic experts said, then I’m very disappointed in their understanding of reality, or their ability to express themselves.

All of which is not to say that 40% of our money disappearing in a little over a year isn’t incredibly disruptive. It certainly is, because we didn’t take away the money evenly. I’ve said it before: if we had everyone take out all their money and add a zero, turning ones into tens, and hundreds into thousands, the effect would probably be unsettling, but not severe — little is changed thereby, and once people adjust to the fact that lunch at McDonalds now runs about $70, we’d be fine. The same would work in reverse: turn dollars into dimes (for everyone) and we’d adjust. But of course this 40% destruction didn’t happen equally for everyone. People lost their houses, their retirement funds, and their jobs. Others didn’t suffer so much, and I’m sure some profited, some wildly so.

*That’s* where the pain comes from.