http://www.choicevalueinnovation.com/

[I kid you not. That's really the official domain of their joint announcement.]

It appears that Yahoo Search will now be powered by Microsoft Bing, and that Yahoo Advertising will be powered by Microsoft adCenter. Microsoft gets to incorporate Yahoo algorithms into Bing (if they choose), and Yahoo is bowing out of the search engine and PPC advertising business. It looks like Yahoo Search will remain, but only as a front-end for Bing. (Though old-timers may remember, Yahoo has been powered by Inktomi and Google at different times in the past.)

Yahoo remains an independent company and will focus on its core competencies (eg Yahoo Mail & Yahoo Messenger, as well as their other web properties, like Flickr).

This deal now gives Microsoft 21.44% of search engine marketshare, based on June 2009 search marketshare figures from Hitwise. They're still only a quarter the size of Google, but that's still enough to make them the #2 player in search, and therefore also in search-related PPC advertising.

So what does this mean for micro ISVs?

• As Barry Schwartz points out over at Search Engine Roundtable, nothing is happening soon. The deal still needs regulatory / anti-trust approval, any integration is unlikely to occur before 2010.
• Longer term, if you have specific search SEO strategies that you use for Yahoo's algorithm, you won't need them much longer. The Yahoo search algorithm is going away. [Shame, I had some good rankings on Yahoo - not that I put any effort into it though.]
• If you have a Yahoo Advertising account, you'll eventually be moved across to Microsoft adCenter. At least you'll be able to handle both Yahoo and Bing from the one advertising account now.
• If you're only working with Google AdWords right now, you'll soon be able to increase your PPC search advertising impressions by another 28.9%, just by signing up for a Microsoft adCenter account. Between those two, you'll reach 95.48% of all US searches.

Of course, one company that is very excited about all this is Ask.com. After years of trying, they've finally moved up a rank to become the #3 search engine ;)

"Do we have a separate QA function at Tapulous? The answer is no. A lot of that is basically because we're a scrappy little startup, and so we are very careful about how we spend our money, and we have limits on headcount. So at the moment we don't have the ability to take on a full time QA person.

Some of that though is also of my Apple pedigree. Although there are dedicated QA people at Apple (and they're fantastic), the burden of initial quality assurance in any given product is definitely very heavily placed on the individual engineer at Apple. I don't believe that's a complete substitute for having a professional Quality Assurance engineer - I would love to have one - but I do think you can do a lot as an engineer to test your own code, test your own products, and just build that into the vibe and the religion of your company.

So that's what we're doing right now. And when we get ready to ship a product like Tap Tap Revenge 2 - one of our big ones - the entire company spends probably a whole day testing the app. We all get together and we bang on it as a team. It's great fun. We find the lion's share of problems, but we don't find them all."

- Jessica Kahn, Director Of Engineering, Tapulous Inc.

Tapulous make Tap Tap Revenge, the most-downloaded game on the iPhone App Store (according to Apple). TTR itself has 10,000,000 unique installs (according to Tapulous), giving it 32% market penetration of all iPhones & iPod Touch devices (according to comScore). They have 15 employees.

There are lots of other terrific hints and tips in the video this is from: "From Upstart to Startup to Grownup: Lessons Learned in the First Year of an iPhone Company" (iTunes link), which you can download for free from iTunes U as part of Stanford University's iPhone Applications Programming course (CS193P). The course videos are available to everyone - you don't need to be a Stanford student or even a registered Apple iPhone developer to download their course materials & videos.

Related Articles

• How To Run A Beta Test... Or Not?
• Should You Run A Beta Test Yourself?

So is this a sign that cloud computing is doomed to failure? That the cloud will never replace the desktop? I mean, Toodledo isn't the only cloud computing company to experience an outage:

• Amazon EC2 had an outage last night after a lightning strike.
• GMail was down for several hours earlier this year.
• GoTo Meeting also had an outage in mid 2008.
• Amazon's S3 Storage took several websites offline during a 2008 outage.
• ... and there have been many more, less high profile cloud outages too.

But cloud computing is a great idea - being able to access your data from any device, anywhere in the world, having your data in sync across all machines... terrific! Problem is, the cloud doesn't solve the problem on its own. Failures do happen. Websites go offline, even for maintenance, and the developers of apps on the cloud need to consider that.

Solving the Cloud Computing problem

The cloud is only an adjunct to a wider solution. A part of the puzzle. A full solution needs to embrace all of these features:

• A cloud (obviously), so there is an online/offsite backup of customer's data, and to facilitate sync between devices.
• A mobile client (eg iPhone), so customers can manage their data while on the go - with a data cache, for when there's no WiFi or phone credit, or if the cloud goes down.
• Desktop client software with a local data cache, so customers can manage their data whether they're online or not. Don't wait for Google Gears - look at how many desktop Twitter clients are available today. Desktop clients are important.
• A web interface (of course), to enable collaboration and for when customers don't have easy access to their own devices.
• An export option from all services (web, desktop/mobile software). If the cloud goes down, they can still export from their desktop. If their computer crashes, they can export from the web.
• The ability to "switch clouds". Sometimes companies go out of business, and their website disappears. Client software should have an option to switch servers/clouds - just like email clients have a settings panel to change SMTP servers.

Toodledo already had some of this in place, like the iPhone app with a local data cache. So I was still able to access and edit my to-do list on my iPod during the outage. But there's no desktop Mac client, and no easy way to export data from the iPhone.

But don't get me wrong - I'm still a fan of Toodledo. Their software is great, they've got the best to-do app available. Being able to access my to-dos on the web and keep them in sync is awesome. Oh, and they've finally come back online as I type this... a quick sync of the Toodledo app on my iPod Touch, and yep, my data is all back to normal. Good job, folks.

But the lesson for cloud companies remains: all Toodledo needs now are good Mac/Windows desktop clients & a way to export the cached data from the desktop & mobile clients... then Toodledo will be the perfect ToDo list solution.

Google Wave does have some interesting ideas - but in my opinion, there's a lot that's wrong with it too.

Google Wave Is A "Single Point Of Failure"

Google Wave encourages you to make everything a "wave". An email is a wave, your tweet is a wave, your blog comments are each Waves that go through a Wave Server.

But several times during the demo, their Wave server crashed. And while the Wave server is down (which inevitably happens to all servers at one point), you can't access your email, and all those blog comment "waves" you posted on your website? Gone.

As things are today, if your email server goes down, at least your website stays online. If your blog software crashes, at least the comments and articles you published stay online. But in the Google Wave world, it appears that if your Wave server goes down, then everything on that Wave Server goes with it temporarily. Emails, websites, internal company documents, they all go offline until the server is fixed. That's a "Single Point Of Failure" - the one thing that, if it fails, takes down everything with it.

They also said that there was always just one copy of a "wave". I assume there are provisions for backing up wave servers, but it isn't something they covered in today's introduction.

Google Wave Encourages Information Overload, Instead Of Solving It

You already have dozens of emails in your inbox. Now imagine every Facebook status update thrown in there as well. Then add every Twitter tweet from everyone you follow as well. Now imagine that even after you've read them, they keep popping back into your inbox because the other person made a small edit. That's Google Wave.

That might be okay if Google Wave gave you a way to deal with it all, and they do provide some tools to stop you drowning in Waves... there are Folders and Tagging options to organize them. But so far, it appears to be something you have to do yourself, by hand.

Compare that with an email client like Eudora 7, that you can setup to automatically sort things for you. It's like a personal assistant who checks your email and tells you which ones are most important. You can create Filters that sort urgent business emails at the top, friends next, and all that anonymous email at the bottom. It can automatically remove the spam and prioritize your email, so you know exactly where to begin with your email each day.

It appears that Google Wave doesn't have that. It even makes it worse, because when you think you've dealt with something, it keeps coming back as others edit it. There was no mention of how Google Wave would handle or prevent spam either.

Google Wave is Hyper-Real-Time

You know that problem you have in MSN Messenger clients, where you're waiting while the screen says "Ethell82F is typing..."?

Yeah, I didn't think it was a "problem" either.

Apparently the Google Wave team does think it's a problem - so they "solved" it by showing everything Ethell82F is typing - as she types it. Typos and all, before she's even finished typing the sentence, let alone the whole message. And they do the same thing with "emails", blog comments... everything. You can watch everything being edited in realtime.

Do you really want to watch 100 trolls all editing their hate comments on TechCrunch, all at once?

I think it's going to be too distracting, watching emails and web pages change as I'm reading them. I'm not convinced that's a benefit yet.

Google Wave is Overly Complicated

Let me quote Lars Rasmussen from today's presentation:

"Imagine you go back and do more work in the original wave, you'll be able to go back and merge just those changes into that same product wave. The product wave will just have a two-step history now, where each step is attributed to whoever did the copy. In fact - and you guys will recognize this as inspired by source control systems - we'll let you have a whole string of work waves, all pointing to the same product wave. You can have different teams working on different parts of the wave, and then when one team merges in their changes to the product wave, the other teams can pull it down, merge it with the local changes and so on."
- Lars Rasmussen, Google Wave Developer Preview at Google I/O 2009

I bet if you say that to any of your non-geek friends, they'll look at you and say "Uhm, but I just wanted to type an email to my friend. How do I do that?" It doesn't have the obvious simplicity of, say, Twitter, or an iPod.

Is Google Fixing Something That Isn't Broken?

There's a whole range of other issues with Google Wave too. I think Dave Collins said it best over on his blog: "Is Google fixing something that isn't broken?".

I wouldn't quite go that far - some of the Google Wave ideas do solve problems. Our modern day inbox is fragmented: there's a Facebook inbox, and a Twitter direct-message inbox, and a MySpace inbox, and an SMS inbox, and that's all in addition to your email inbox. There is no one application you can use to reply to them all in one place. (Though you can at least receive most of those in one place - via email.)

There are already companies working on this problem. Trillian and Adium solved it for instant messaging. EventBox is solving it for social networks, and PostBox is trying to do it for email & social networks. GrandCentral is doing it for email & phones & text messages - and that's another Google project!

There's still work to be done - none of those applications appear to have auto-categorizing and auto-prioritization. But they're closer to the goal of fixing email than Google Wave is.

Google Wave is a neat Wiki/CMS/collaboration platform - but it isn't the email replacement they claim it to be. I'm not sure why they thought a Wiki is the right solution for email. They're asking the right question, but they've got the wrong answer.

I've never tried outsourcing beta testing to a dedicated service, so perhaps there's something I've missed out on. But there are several advantages to beta testing directly with end-users / customers that I don't think a beta testing service can replicate.

Before I continue, a caveat: this is written from the perspective of a small software startup developing consumer software - a microISV with minimal funding. It may not apply for a large software company writing bespoke or B2B software.

Testing on diverse platforms and real world systems

Obviously, we need to test our software on every version of Windows, and any decent beta service will offer this. There may be some doubts about whether they have older systems - eg Windows 98 may be obsolete, but a small percentage of our customers still use it, so we still do some testing on it. I assume that a beta testing service would be able to do that for us too.

Testing on multiple Windows versions is something we already do in-house though. With an MSDN Operating Systems subscription and a copy of Virtual PC, Parallels Desktop or VMware Fusion, it's easy to test your software yourself on multiple versions of Windows - even if you only have one PC. Honestly, this is something any serious developer should be doing already. (And we actually test all the way back to Windows 95a - you'd be surprised at the bugs that can be revealed by testing on those old systems!)

But the one thing developers can't test on - and that I don't think beta services can either - is a real world customer system. Some customers have software installed that conflicts with each other, has esoteric settings, or might even be junked up with viruses. By running a beta test with our customers, we see how our software works on systems that real customers will be using - our clean VM installations of Windows aren't accurate representations of that.

Would a beta testing service have all the software you need?

Another issue (for our software, at least) is whether professional testing companies have access to all the necessary software. In our case, that would mean every version of Photoshop from version 6 through to CS4, every version of Paint Shop Pro from version 5 to X2, versions of Photoimpact from v8 onwards, copies of Photoshop Elements, and so on.

We have access to many of those programs through our relationships with those companies. For some programs that we don't have, our customers use them and can do some testing for us and provide feedback. But for a professional service to buy those licences would cost several thousand dollars, and some of those programs are no longer available for purchase anyway (except, perhaps, secondhand via eBay).

This is probably something unique to our software, though. If your program is a more mainstream out-of-the-box application, it might not be a concern for you.

Customer beta tests help you learn about your market

As part of our beta-testing signup process, we ask for a lot of information. We do this to screen for serious/genuine applicants, and to make sure we get a wide cross-section of users when testing (eg making sure we have people of all ages and genders in our beta testing). But from the signup applicants, we also get a sense of who is interested in our product - we might see more of one age group or gender than another. By talking with our customers during the test, we learn more about their interests, the websites they visit, and their needs for the software. We can also see from the number of people signing up whether a product is likely to be popular or not. If people aren't signing up for the beta, then maybe we need to tweak our marketing messages or reconsider the project.

Beta testing with customers helps you launch your product

On one occasion when we worked with our beta testers, not only did they get a free copy, they were also given a limited-time discount coupon to give to friends. That created significant word of mouth and was a key factor in the launch of the product. While you can approach bloggers to talk about your product, people are more likely to talk about your product if they're a part of its creation. And our beta testers should feel that way, because the testing is a significant part of our development & design process. We try to credit all our testers in the software, as a recognition of that.

Big companies don't use beta testing services (or do they?)

Many indie developers have been involved in private beta testing for large commercial software companies before. We know that they work directly with their customers for beta testing purposes. Perhaps they're working with beta services as well and we just don't know - in fact, they probably are. But if that's the case, and the private beta testers are getting the software after that outside testing, then the bugginess of the software we received doesn't say much for those services!

In all honesty, I don't know this for certain. What I do know is that amongst smaller startup developers, it's rare to hear of them outsourcing to a beta testing service. Of the larger companies I've worked with, it was never disclosed to me if an outside group was also involved, but that's understandable.

Can a startup / microISV afford a beta testing service?

This is probably the biggest issue in the way of a startup software company using a professional beta testing service.

Many of the beta testing services that I've seen provide no pricing on their website. While that's totally understandable, there's a general rule of thumb that startups follow - if the price isn't listed, you probably can't afford it. An indication of costs (without having to "call for pricing") would help startups / microISVs make that decision early on. Without that info, they'll often just ignore the business.

Anyway - none of this is meant to be negative against professional beta testing services. I've never used outsourced testing services personally, and maybe they're quite useful. If you're having difficulty finding beta testers for your product, perhaps it's an ideal service for you. But I don't think an outsourced beta testing service can completely replace the benefits of beta testing directly with customers.