(this is page 110 of 121)





Latest

Making the Hard Calls on APIs

As usual our friends over at TechCrunch have ignited a debate around web services, data, and the openness of APIs. This one centers around the fact that Zoomr, a Flickr clone, has been denied access to a commercial API key to export people's photos/metadata out of Flickr and into Zoomr. Since we've been mentioned a few times, and since the issue at the core of the debate (should Flickr allow a direct clone access to the data it has built up despite the clone's desire to "replace" Flickr) is a tricky one, I thought I'd give some quick impressions.

First, to the importance of official "openness" in an API: when we were doing early alpha testing, there were two applications that our lead users told us we had to integrate with: Flickr and Picasa. Flickr was no problem: they have a process for applying for a developer key, a clean and well-documented API, and a bunch of third-party libraries/apps that you can use as examples. Picasa? Well, that was a bit trickier. No documentation, no API, and no seeming desire to do anything about it. Fortunately for us, we have someone at Tabblo who relishes these kinds of challenges, and so with a couple of days work, he had it reversed-engineered and we were off and importing photos from Picasa.

So my first point is this: if it's important enough to your product/service to get at some of this stuff, you will find a way whether the company or service you are dealing with is "open," "a model Web 2.0 citizen," or not. Sure they may be a little more friction involved, but one of the great things about networked services is that at some point you have to get on the wire and push bytes, and when you do, it is not that hard for someone else to come along and learn how to speak your language.

Second, and this is the tricky bit to grok, when you use an API like Flickr's, the goal should always be to benefit the user and not to "jumpstart your userbase" or "blow out the metrics." In our case, we wanted to be able to integrate with Flickr for two clear reasons: 1. the avoid the user the hassle of re-uploading photos that they wanted to use in their tabblos, and 2. to avoid the user the hassle of re-organizing the photos with tags, dates, etc. But as far as where the data lived, the stickiness of the app, etc., we really didn't care (in fact, at first we wanted just to use thumbnails and metadata and leave all of the original file-serving on Flickr, something which we had to give up on only because it impacted the user experience.

And since this is a clear enough goal, I think that it should be ok for the person doling out the keys to the API to make a judgement call about whether you are in fact trying to do something that truly is going to benefit their users with the API. I realize that defining this is tough but just because the line in the sand in hard to draw, doesn't mean we shouldn't try. In my mind it has something to do with the nature of the features/functionality being offered by the API requestor, and whether there is a clear and demonstrated need on the part of the community of users for that sort of thing. At Tabblo, we talked to quite a few Flickr users who said things like "well I'd never leave Flickr as my primary place for photo uploading/storage, but what I'd really like to be able to do is ..." and it was the "..." that we listened to really closely.

Now the Zoomr guy may actually have a pretty good "..." to offer Flickr users, and if he does it may be the wrong thing to deny him the API key. But here is where I fall back to my first point– if it's really all that critical to get those Flickr folks in, just do what we did with Picasa and figure out how to get a minimal export working. If enough people start doing it, and given how well Flickr listens to its community, it won't be long before you get the keys to the data. Generally speaking the great thing about clued in companies born of the web (Yahoo, Google) unlike those born of the PC era (Apple, Microsoft) is that they listen and adapt quickly.

Of course this doesn't answer a lot of the hard questions that people are asking with respect to data storage on the cloud in this brave new world but unfortunately we are all ankle-biters when it comes to the really big questions here, and only the big guys can help make the right policy decisions about what should happen in the grand scheme. As more people from the clued-in companies climb the ladder at those places though, I have faith they will influence things in the right direction. Until then, we should all cut them a little slack.

Comments Permalink

Google gets in the game

Rafe, who I am seeing today at the Under the Radar conference, broke the news this morning that Google has finally launched their first foray into online photo sharing, Picasa Web Albums.

The implementation looks to be a really nice version of a clean photo-gallery plus a "photo-cast" capability that allows other Picasa users to download the photos that you publish. It's clear that what Google is hoping to do is leverage their gorgeous Picasa client to give people a way to begin sharing their photos online in a simple and basic process that doesn't allow more much beyond basic sharing (this is probably made obvious by the their own blog post titled: "It’s all about the photos.")

I'd be worried though, that without more advanced features specifically around privacy, presentation, and collaboration, the product may fail to attract the increasingly more sophisticated publishers of photos online (then again, the product may be geared towards the 30% of folks who have never shared a photo online). For instance, most of us have loads of different groups that we communicate with online and a simple two-state privacy system (private/public)– while better than the all-public model of most blogging engines– doesn't really begin to address the ways in which we'd like to represent ourselves to others via our photographs.

More importantly, I want much more creative control than the standard grid of photos plus zoom-in view that Picasa Web Albums provides. It's interesting to me that most of the online photo sites copied their interface for display (the standard grid) from desktop photo management software even though that interface was built mainly for organization/editing and not consumption. It's almost like if magazines from the newsstand were laid out like the Quark or InDesign interface. The grid is easy for sure– it requires very little creative investment on the part of the user sharing the pictures– but as anyone who has ever been the recipient of a "weddingstorm" or "babystorm" can tell you (this is what happens when your friends start getting married/having babies and discover how easy it is to spam everyone repeatedly with huge online gallery links), after the first 9 grids, it starts to get old. With all of Google's AJAX goodness, I was sort of hoping we'd get a lot more of the cinematic effects that Picasa became famous for around the composition and story-telling aspects of photo presentation and display.

Finally, I wanted Picasa web albums to be more like Google Spreadsheets in the way that the latter breaks new ground around collaboration. As one would expect, the blogosphere is already abuzz with posts about how this is the first volley in a showdown between Picasa Web Albums (Google) and Flickr (Yahoo), but I couldn't disagree more. Flickr is the world's best collaborative digital shoebox with a few people throwing stuff in, and many many more sifting through the box for the best photos, helping to reorganize them, and writing small comments on the back of the pictures for others to read. It works so well, the experience is downright entertaining. But Flickr is not a photo editing tool, a photo organizing tool, or (as mentioned above) a photo compositing and layout tool. Picasa Web Albums, on the other hand, is an online extension of a rich client tool that does do some of those things. Nick Starr wrote a few days ago that Google was trying to build the new Dot Mac with tools that allow for easily publishing content that resides on your PC, and to this end, Picasa Web Albums is a very nice first step. But to me, collaboration around photos could be a lot richer than that, and we'll have to see whether it gets there as the team evolves the product.

Comments Permalink

Note: this post will be of little interest to you unless its title grips you like a Richard Morgan novel (ok, John Grisham for the non sci-fi fans) and you are in the middle of tearing your hair out trying to get EVDO working on your MacBook Pro (MBP). Read on if appropriate...

I got addicted to Verizon's 3G EVDO network back when I was riding the Acela between New York and Boston twice weekly. With it, you can get real work done on a laptop, and it tends to be rock solid when you can get a connection. However after releasing very beta hardware in the form of the early MacBook Pros, Apple also nuked the PCMCIA slot in favor of an Express34 one which meant that my Novatel 5220 card was left a paperweight (an Express 34 EVDO card is "just around the corner" apparently).

Enter the Treo 700p to the rescue. It supports EVDO natively, doesn't run the slow-as-molasses mobile Windoze that its predecessor did (the 700w), and best of all can be configured to use DUN (dial up networking) via Bluetooth with the MBP. Now, folks have complained that it is not possible to get the MBP to work with the Treo as a USB EVDO modem because this is supposedly faster than Bluetooth while keeping the phone on trickle charge, but based on a few tests I ran (the MBP/Treo versus a Powerbook G4/Novatel 5220), the speed difference amounts to line noise around 25-75 kbps on download and 10-15kpbs on upload. And if fact, I find the notion of having to whip out a wire on a train/airplane/cab to "tether" the two devices sort of stupid (isn't "tethering" something we do to horses?).

Anyhow, without further ado, here are the instructions to get this done with minimal third party software getting installed on your computer, and the least amount of hunting through forum posts. I'm doing this because the first pairing of a cellphone and a Mac I did was greatly helped by a guy who took the time to do this with a v710 a couple of years ago, and I got many miles out of it. Oh, and I am also doing it because the Verizon support reps. are absolutely clueless about Macs, Broadband Access on EVDO, and just about anything you may want to get done that doesn't entail accessing voicemail. No joke, and no rant intended, though it does amaze me that they can be so bad at selling what is one of their highest margin and best offerings.

Before you start

As of 6/10/2006, Verizon has a couple of plans that provide unlimited data on top of some allotment of minutes. Go and sign up for this, but make sure that you also sign up for a $15/month "feature" to let you do this in the first place called something like "Broadband Tethered Access." Just check that a) it is $15/month incremental for nothing other than the privilege of using your Treo as a modem, and b) that when you log into your account and check your service, you see a line item that reads like so: "BBA CONNECT UNL." Without this enabled, you will not be able to authenticate. In my case, I asked for it at the store (didn't get it), and twice on the phone (didn't get it). The trick was getting to the data support people and using the magic words "BBA CONNECT UNL" which seemed to turn some lightbulb on.

Also, you do not need to install the Verizon Access connect software that they put out for the Mac. I am not sure whether this software is good or bad- in fact I've only ever used it on my Thinkpad where the first thing it did was flash my otherwise very Mac-friendly 5220 so that it only worked just ok after that- but I prefer to keep as much of that third-party crap off of my computer, particularly when it comes to OS services like networking (and given that the MBP seems to have a few hiccups of it's own in this area, why add insult to injury?).

Getting the Treo and the Mac friendly with each other

I didn't install any of the Palm Desktop manager software, mostly because I was afraid that it would run in emulation and really clog the MBP up. Instead I got a $39 program called the Missing Sync which I'd heard was great and had an Intel binary. Either way though, pairing your MBP and your Treo should be a simple matter of going to the Bluetooth control panel in System Preferences and adding the Treo as a device. Make sure you turn Bluetooth on on the Treo itself (mine came off). You will go through a standard wizard on the computer where you have to enter some numbers the MBP gives you on the Treo and get stuff paired correctly. The key thing to realize is that the process is initiated from the MBP (not the Treo).

Once you finish the pairing (assuming it is successful), the wizard will ask you if you want to use your phone for Internet access. Choose yes, and enter the following bits of information:


Username: XXXXXXXXXX@vzw3g.com (where the Xs is your 10 digit Verizon phone)
Telephone Number: #777
Password: vzw
and click the box for "Show modem status in menu bar."



If you've done this all correctly, all you have after you've closed everything out is pull down on the little handset icon that appears in the menu bar and choose "Connect." The computer will display the various steps in the process and the Treo's screen will go on to indicate that the process is going.

When things go oops

First, make sure you've signed up for the "BBA CONNECT UNL" service option. Without it you are hosed (unless there is some modem init hack that you can do). If you see the service on your web-based Verizon portal, make sure that Bluetook pairing between the MBP and the Treo is solid, deleting other Bluetooth pairings if necessary.

Some things I've noticed

The connection speed is comparable to the old PCMIA setup (350/60 kbps on dslreports speedtest versus 400/70kbps) but from very early testing it seems to drop after a smaller interval of inactivity (I'm sure this is configurable but I haven't had enough time to explore). Apparently others have written that with this setup an incoming voice call will allow you to pause the EVDO connection without having to disconnect (a big improvement over the pre-EVDO days) but I haven't seen this first hand.

That's about it. I hope someone gets to save some time in reading this the way that I did a few years ago. After all it is only by doing write-ups like this that we customers are ever going to be able to build a decent support system around Verizon's great network. Now if we could only do this to build actual network infrastructure as well...

Comments Permalink

Chris Anderson of Long Tail fame has a couple of posts on Lego Factory, a "CAD/CAM plus mass customization" product line that is fascinating to see, and even more enjoyable to actually use.

Basically Lego Factory works like this: you download a free design tool from the Lego website and start building your own custom Lego model in what is clearly a simplified CAD tool. You work from a palette of standard Lego pieces but get to choose one-off from any of the available pieces (including those ones that annoyingly only include three to a kit when you need four). Since the app is available on both Windows and Mac, I was afraid it was going to be some kludgey Java thing that was frustrating and limited, but it's actually a native binary that is snappy, intuitive, and attains a great balance between power and simplicity (and this is on the Mac no less).

When you are done assembling your model, you can price it on the website and order it as a kit-of-one made especially for you. Anderson's second blog post covers how back at Lego HQ each kit gets picked, packed, and shipped by manually and talks about the seemingly unbecoming economics of this but based on my own experience, I think that this is something they will quickly sort out as demand ramps.

I'm an absolute Lego nut with a four year-old that serves as a terrific enabler when it comes to sinking time into things like Lego Factory but Factory is interesting to me from the perspective of understanding the coming phenomenon of mass customization in consumer products. Levi's took on the first great experiment in this a few years ago with "jeans that fit every butt" but since then I've been surprised by how few people have followed suit. Recently, someone sent me a link to an Beyond, outdoor clothing manufacturer that is taking a similar approach with clothing. For years I've been yapping to anyone who will listen about the need for a custom bag equivalent of Beyond clothing, and yet no one seems to be ready to answer the call even among a slew of innovative, high-cost producers of bags.

I think that folks who look into these markets are afraid of two things: 1. as Anderson points out, at first blush the economics smell really bad (especially when you think about competition from China) and 2. I think that folks are afraid that, as Levi's "discovered," the demand is simply not there at "mass market scales."

I think that bad economics is a bit of a red herring from a couple of reasons. First, again from my own experience, it's amazing to see how cheap these "high touch" custom products can really get if you apply even a basic amount of engineering to the manufacturing process for a few different reasons (to come in a latter blog post). And more importantly, a growing group of consumers have demonstrated a clear willingness to bear higher prices for customized goods.

And as far as demand goes, you need to look no further than Anderson's own forthcoming Long Tail book to see how the Internet can be used to aggregate enough niches to get to pretty significant numbers (I've been reading an ARC copy of the book and have to say that it is an awesome read even if you feel over-exposed to the all of the long tail hype). Besides, the Internet's enabling of peer production is the rocket fuel that mass customization needs to really explode as a far-reaching phenomenon. Lego Factory is beginning to hint at this through the community and contests, trying to encourage users to leverage each other's work, but I want it to get both smaller and bigger at the same time. Smaller because I want to learn how to assemble wings, claws, etc. from the lead users that can make New York skylines with full Statue of Liberty replicas- I may even want these gurus designing custom pieces that I can buy for my kits. And bigger because I also want the community around Factory to thrive in other ways: tutorials, videos, and perhaps even live seminars that allow those of us struggling to get what we want to be able to learn "guild-style" as we might have 500 years ago by apprenticing with an artisan.

It's clear that the computer, cheap tools, user-generated content, and the Internet have busted open the media landscape for good. Now I not only consume when I want to (a la Tivo), but also where I want to (iPod ecosystem), and most importantly what I want to (a la YouTube, podcasts, etc.). I want to see the same thing happen with physical products across all sorts of categories. We're trying to do it in story-telling products, but others, like Lego with Factory, need to get on the bandwagon and bring us closer to a world where atoms start behaving just like bits are starting to in media.

Comments Permalink

Google Spreadsheets is Great

I have to say that I am really impressed with Google spreadsheets. It's the missing piece to moving entirely off of Office as a tool for coordinating work for us. In my world, the wiki killed Word but it's always had this horrible problem with tabular information (or any sort of structured information for that matter). Google Spreadsheets takes care of that AND adds realtime multi-author capabilities. Dan Bricklin spotted this need in presenting tabular information on the web months back which is why he got going with WikiCalc. And despite the fact that I'm not sure I'd want my company's product schedules living on some Google Labs server, I think most people won't care, and for the 80% of users who use Excel as a fancy table editor, this product is going to really hit the spot.

My only minor complaint is around the lack of Safari support. Safari sucks as a web development environment-- there is no doubt about that, but if even small companies can come out of the gate supporting it, why can't Google plan for a degraded experience in Safari from day 1? Otherwise, Spreadsheets does a really nice job of providing the basics around editing as well as an amazing job of Excel file import (haven't tested the export).

Now, it would seem in a predictable twist of fate that Google has run out of good graces in the blogosphere. Everyone seems to be piling on with respect to how ho-hum this is, or how non-core to the mission it is, or just how much Google sucks now. Then there are the people who claim that this will never be able to replace the richness of the Excel experience. Having recently seen a demo of Excel 2007, I'd have to say: yeah, but so what? Outside of some formatting goodness which Excel 2007 seemed to have in it (which could be done in an AJAX GUI), most of the other improvements didn't strike me as particularly useful.

Besides, I think that Nick Carr has it right when he writes about how Spreadsheets should be seen as complementary to Excel (and I don't usually agree with his shrill blogging). If you buy the complements argument, which seems to have a lot of legs when you think of iPhoto/Flickr or iTunes/LastFM where each of the online services complements the rich client application with publishing/sharing/collaboration featuresets, then Spreadsheets is an absolute no brainer. In fact, it could do a lot less than it already does and still be quite useful as a complement to Excel.

Comments Permalink