Digital Earth Weblog

Sydney was the first Google Developer Day 2007 event out of sixteen that were held around the globe on 31 May 2007. With seven hundred attendees it was also the second largest event, exceeded only by the final event at the Googleplex in Mountain View, CA. It was a good day out.

The entire day was very well run, from the lovely ladies who greeted us at the door with a smile, to the presentations, to the after event drinks. James O’Loughlin (ABC Radio and The New Inventors) did a great job as master of ceremonies. The seven huge screens, thumping sound system, and professional audio visual crew certainly helped. We were all fed in pretty quick time, and there were quite a few interesting displays to look at. Plus, of course, free wifi and charging stations.

However, the standout factor was the Google people themselves. The red shirts. Those that I spoke to were very friendly and helpful. The problem was that there wasn’t enough of them to go around during the breaks. There were small queues of people waiting to talk to some of them.

With 700 people in the audience it’s difficult to open up the floor to questions. That meant that the information flow was one way during the presentation, with a passive audience (not my favourite way to acquire information). No doubt this increased the demand for red shirts during the breaks.

Feedback on GDD07

Given that it was a developer day I was expecting two angles:

1. Some brief marketing on why developers should use Google tools.
2. How to make those tools really sing. e.g. Tips and tricks.

There was some low key marketing, which was fine, but some of the presentations weren’t as advanced as expected. It’s more useful to know what can be done with the technology, and what the key factors are in making it work. Showing people KML samples is fine, but we can all view source. The stuff presented on KML regions was at the level I’d like to see more of.

The only technical issue was that the two screens in the centre used to display presentations and code samples weren’t up to scratch. I think they were rear projection screens, and they didn’t have the resolution or contrast required to display small text. Luckily, the two screens on either end were very readable from a distance.

Suggestions for GDD08

Some quick thoughts:

• More mini case studies. Who, why, what and how. Inspire us.
• More technical sessions where people can ask questions.
• Promote partner programs (when available in Australia).

NB: I realise that the event only goes for one day, and that the Sydney office only has staff working on a few Google projects. Perhaps more international guests would be required.

Course is over. I’ve got to say that it was one of the most enjoyable courses I’ve ever been on. The instructors and the participants were all passionate about their work. All the participants were there because they wanted to learn how to improve their software development processes.

Apart from the core Scrum discussions there was also a lot team exercises. Both the instructors and participants provided many case studies, and discussion ensued. Basically, it was fun, the way software development should be. The day ended with a team photo at a nearby pub, and few pints.

For a hint at what this lean/agile development is all about check out the Agile Manifesto.

The first day of this two day course was very interesting and informative. Jens and Dave, our instructors, are full of wisdom and interesting stories from companies using Scrum.

The key take-away: Scrum is simple, but implementing it is hard.

The main reason for this is that it shines a big bright spotlight on software engineering practices and organisational issues (as do other agile methodologies). However, those organisations that get it right are, roughly, doubling their productivity.

Australian interest: If you are interested in agile software development processes there is some Scrum Master Certification training happening on the east coast in March and April 2006.

There are quite a few agile software development processes these days. Extreme Programming (XP), Crystal Clear, Lean Development, Scrum, etc. (Martin Fowler has a good overview.) All agile processes focus delivering business value as early as possible. Scrum appeals to me because it is relatively a simple methodology that:

• Gives the customer the responsibility for prioritising the work.
• Gives the development team the sole responsibility of estimating effort and delivering product.
• Uses iterative development in which each iteration produces high quality, tested, documented, working code, which.
• Reduces the cost of changing requirements and priorities.
• Encourages cross-functional self organising teams.

The 2 day Scrum Master Certification courses planned for Australia are on the following dates:

• Sydney – 30/31 March 2006
• Melbourne – 3/4 April 2006
• Canberra – 10/11 April 2006

A good place to learn more about Scrum is the Control Chaos website.

Seth Godin has a timely post on How to Run a Useless Conference. I’m not sure that I total agree with his statement that “Facts don

It was an interesting day today. I attended a morning session that covered some more case studies. Some very good work has been done. The middle of the day was all Jack Dangermond, who judging by the crowd’s reaction lived up to expectations. The afternoon was all ESRI Australia doing their thing.

The morning session that I attended was focused mostly on data management issues. The talk was about geodatabases, data models, and ETL (extraction, transformation and loading). A few presenters even got into some detail about what they had done, which pleased my geek side. (Most of the presentations were focused on “business benefits”, not the details.)

Strangely, tacked onto this session was Sean Skilton from Sun who went through another presentation on service oriented architecture (SOA). I thought his explanation of what SOA is was quite good. However, as before, there was no details given on how you migrate to a service oreinted architecture, or what some of the issues are. The big services companies obviously want to do that for you, and have you pay them to do so.

All that conference attendees got was that that the spaghetti created by N to M web service connections get moved to a new layer in the architecture (which others referred to as the Enterprise Services Bus) which helps exploit and manage those services without having to write a lot of code. The principle is sound, but no path was offered. Just apply magic vendor glue product and it will all be solved.

I spoke to some of my clients afterwards and they were still not clear on the whole SOA thing. The message seems to be getting lost in a maze of buzzwords.

Jack Dangermond was a breath of fresh air. I had never seen him speak live until today. His passion is obvious. He did a great job of explaining his vision, and articulating where ESRI products fit into that vision (something that had been lacking at the conference). He also had a gentle dig at a few of his own products, which was refreshing after hearing nothing negative about any ESRI product during the rest of the conference. They can’t all be perfect, otherwise we wouldn’t need all the great new things and enhancements in 9.2. I got the impression that he would have liked the ESRI product suite to be a little closer to his vision than it is (that’s my personal perception, your reality may vary).

Yes, ArcGIS Explorer was demonstrated, live. It’s splash screen was labelled “E2”, and Jack Dangermond suggested that it’s name has not been finalised. It looks very good, and Jack mentioned both WMS and WFS (no details given though).

Personally, I’d like to see some of the more gimicky predefined searches disappear. The left hand panel of the user interface is quite cluttered. For example, “Where’s this computer?”. Best case scenario: This will get used once and give an accurate result (it was close when Jack used it, but Brisbane is not Main Beach on the Gold Coast). The worst case is that it will get used once and return a false result.

I’m keen to get a beta copy of ArcGIS Explorer. It shows a lot of promise, and I’d be happy to provide some feedback like I did with Google Earth. As always, data will be what makes or breaks it, and it looks like a lot of the responsibility data serving will left to organisations that have ArcGIS Server. Although, I’m not clear in exactly how this will work other than there is some mechanism to prepare data prior to it being served. It sounds a bit like pre-rendering image tiles to get performance.

The afternoon session was ESRI Australia doing education sessions. I attended the enterprise GIS session and I found it to be 50% sales pitch. This was strange for what I thought was a users conference, so I skipped the last session and went to get some fresh air.

When I got back to my apartment I read the fine print and found that OZRI is formally a “National Client Conference”, not a “User Conference”. I’d agree with that.

So, it’s all over for this year. Would I go again? Hmm… probably not for a few years if it stays in the current format. If it changes to a true users conference I’d be much more interested because I got very little out of it at a technical level. It was good hear what the big vendors are telling their clients, and to catch up with people I hadn’t seen for a while.

All up, when I consider the cost of the conference, the cost of travel and the lost work time, I’d say that I didn’t break even. But hey, that’s just my take on it.

Update: Just to be clear, my company (Digital Earth Pty Ltd) is not an ESRI client. However, quite a few of my clients use ESRI technology. My OZRI posts are written from the point of view of an outsider looking in. Your mileage may vary.

The “enterprise GIS” session after lunch was mixed bag. Instead of going to the final session, which was basically ESRI Australia doing their thing, I spoke with a few of the vendors at their booths. Interestingly, of the four presentations given in the afternoon session the two most informative were from ESRI business partners IBM and Information Builders, not ESRI clients.

Kym Farnick from IBM actually gave definitions of “service” and “service oriented architecture” that mere mortals could understand. He did a good job of making the whole web services stack understandable, and didn’t harp on about IBM products.

One telling point in the IBM presentation was where ESRI tools fit into the IBM web services stack. It is a very small component of the overall stack. The implication was that ESRI plugs into IBM, not the other way around. That certainly makes sense from a business IT perspective, and ESRI have stated their desire to own the geospatial block in the SOA diagram.

However, in my view ESRI Australia’s communication of what they think SOA is, and how their products fit into it, left a lot to be desired. I’m sure there were a lot of people in the audience who didn’t have a clue what it was all about. I certainly didn’t get a good picture of what they are doing.

Bob Hazelton, from Information Builders, did a good job of showing how disparate information sources can be leveraged and turned into web services without writing code. He had some interesting examples of how adding a geospatial component to business intelligence applications has achieved significant results in the US. The classic “build vs. buy” decision hasn’t gone away, it has just moved up the stack.

None of the examples were Australian, although IBM have been doing something at the NSW Dept of Lands. It seems we’re a bit behind the curve on the use of web services in Australia. Perhaps we don’t have environments of the scale that justifies the use of SOA approaches.

On the whole the audience has been quite passive. Only one presenter has offered any “contructive criticism”. No questions have come from the audience. It’s quite weird, well, to me anyway. Sort of like an audience watching a TV show.

On blogging, I’ve not noticed anyone else tapping away on a laptop.

OZRI 2005 is smaller than I expected. There are only about 300 people here. Still, it was good to see a lot of familiar faces.

The presentations from the morning session covered a lot of ground. Brett Bundock (CEO ESRI Australia) showed some interesting statistics in general industry trends. Virtual Earth, Google Earth, and open source software were all mentioned briefly, as was service oriented architecture (often).

The statistics were sourced from various locations (mostly US-centric). One interesting statistic on use of the Internet was that 84% of people use search engines to find information. No surprises there, but apparently the same percentage of people are searching for locations and driving directions.

Service oriented architecture: SOA got mentioned quite a bit and several of the case studies, but no solid definition of this term was given. Web services are rightly seen as the mechanism that will make spatial applications part of mainstream IT. Buzzwords like “Orchestration” and “Enterprise Services Bus” were also mentioned. This reflects ESRI’s push into mainstream IT (more on that later).

Web services: Web services seem to be a coming to the fore in the enterprise space. However, several presentations this morning showed that there is still a teminology issue that is confusing that conversation. There are the web services that are “software systems designed to support interoperable machine-to-machine interaction over a network”, and there are web services that are “services delivered to the user community over the Web”. With all the talk about portals this morning I’m sure that the web services message was not clear.

Open source software: Apparently Gartner is predicting that open source software will not exceed 10% penetration in for Global 2000 (I didn’t get what period that coveed). The important point that ESRI Australia chose to make from that was that revenue opportunities are shifting from licences to services. Pay per use models are becoming more appealing, and web services can play a part in that.

Case studies: Four presentations were case studies highlighting business benefits, all using ESRI software, of course. Some good work has been done.

Unfortunately, Jack Dangermond has been delayed and will not be here until tomorrow. Oh, and no mention of ArcGIS Explorer yet. I assume that has been saved for Jack.

I’m off to OZRI next week. It has been quite a few years since I attended a vendor conference. I’m ready to be impressed, but I gotta tell ya, the bullshit filter has been tested and found to be operational. I tested it with “that’s coming in the next release” and it redlined. Check. :-)

The availability of wi-fi at the conference is an unknown, but I’ve got the EV-DO wireless card as backup. Having never blogged from a conference before I’m not sure how it will go. Don’t expect it to be “live”, or even close to real-time. I’ll try to pick out the interesting bits.

I’ll be wandering around in a white polo shirt with the Digital Earth logo on it (see right column). Feel free to say hello.

Note to self: Hurry up and put a photo of your ugly mug on this site.

The SIDP Roadshow hit Sydney on Tuesday morning, and it exceeded my expectations, but not by a lot. The overall message, of which I’m still a little dubious, was that the technical standards required for spatial interoperability are now stable, well understood and supported by a range of vendors including the big names in GIS. However, there are still business level issues (hey, people are involved so no surprises there) and no real progress on a shared catalogue service.

All the usual suspects were there, but it was encouraging to see some new faces from both the public and private sector. No doubt this was helped by the three interoperability scenarios being focused on emergency management and insurance issues, rather than a natural resource bias. A rough head count suggested that the vendors almost outnumbered the non-vendor participants.

It was a very pleasant change to see four vendors sitting beside each other (Earth Resource Mapping, ESRI, MapInfo and NGIS), and doing live demonstrations using web services located all over Australia and one in Canada. Apart from a bit of Internet latency and server delay the demonstrations went smoothly. Well done people!

Most of the demonstrations were done using Web Map Services (WMS), which is quite a mature standard, but there was also some Web Feature Server (WFS) action, which was good to see. It would have been interesting to know where the inevitable technical issues arose, what didn’t work as advertised, and whether it has improved since my days supporting the CANRI framework. However, even the technical section of the programme was aimed at a non-technical audience. It got about as deep as “your stuff speaks the same gibberish as our stuff” (quote from Dr Lesley Wyborn at Geoscience Australia) so I didn’t both asking any technical questions.

The number of services available on the Internet is growing, and commercial data brokers are getting involved. However, finding both free and commercial services is still a big challenge. There is still no single well known catalogue in Australia that you can query. Given the number of questions about how to find services that is an issue that needs to be resolved.

Speaking of data, it is still a major issue. The usual questions arose about copyright, quality, pricing, liability and access.

Overall, the message being delivered was that the technology is maturing so the conversation can move “up the stack” from the technical issue to the business issues.