Software complexity has been steadily on the rise. Some persistent problems
have dogged programmers for 60 years, and still need solutions. New problems
arise as the landscape changes. What are the opportunities and maintenance
challenges ahead as teams work faster and collaborate across different time
zones, languages, and countries? What areas might be investable? This panel
of experts will explore these questions and new developments in a spirited
discussion on the state of software development.

Notes and Comments by Barney Pell

This document contains my notes taken as an audience member attending this
panel discussion. While it might look like a transcript, and I attempt to
capture everything in real-time, I do also interpret, paraphrase and
summarize as I type.

Personally, I thought this was a great discussion, and one that should be
shared with entrepreneurs, investors, vendors and customers in the software
development community. I worked hard to make this article readable so that
others can benefit.

Highlights and take-aways:

• John Mashley pointed out that, independent of all the improvements in
  software development tools, individual programmer productivity is always
  nonuniform. Some programmers are 10 or even 100 times as productive as
  others.

• This has negative and positive implications for outsourcing. On the
  negative side, if you can assemble a crack team of programmers locally who
  are 10x more productive than you can find offshore, then you actually
  lose economic value by outsourcing. On the positive side, some of the
  best developers are not local. So despite the hassle, and even if the
  labor rates are not much different, you can gain economic productivity by
  taking advantage of these fine programmers wherever they live (whether in
  India or on a boat in the Carribean).

• While outsourcing is increasingly becoming an imperative for businesses,
  the results of outsourcing are often disappointing. Steve Mezak discussed 7 major
  categories of mistakes in outsourcing and listed potential solutions
  (including methodologies and tools) for each.

• A recurring theme in the panel was the difficulty and importance of
  getting software requirements right. Despite all the improvements in tools
  and processees for developers, there have been limited improvements in the
  way people create and validate the requirements in the first place. Errors
  in requirements ripple through the downstream flow and get more expensive
  to fix the later they are caught. N8′s David Hartford brought up the
  statistic that requirements errors cost US companies $100 Billion per
  year in rework and cancelled projects alone. Steve Mezak listed a new
  generation of companies (including N8) that address different aspects of the
  requirements problem in different ways. I am very familiar with N8′s
  Scenario product that David discussed and demonstrated at the meeting. I
  think it has the potential to address the requirements engineering problem
  in a substantial way for the first time. (I intend to write more about
  this in a separate post).

• There was consensus that distributed development is now working. This has
  been one of the biggest problems with outsourcing, telecommuting, and
  large scale projects. There was discussion that open source tools have
  matured substantially and are now being integrated into cost-effective
  suites. In addition, VoIP (e.g. Skype) has enabled people to maintain open
  voice channels all day long at low cost. As Sam Jadallah said:

  I have a company with 2 people, on in Silicon Valley, one in
  Dublin. Skype sits open all day long, just like two people sitting in
  room together all day long talking whenever they want. The tools,
  workflow, process, and experience level is now working.

• Sam Jadallah’s discussion of VC investment in software development tools
  is great reading. His fundamental point is that the internet has changed
  the economics of the enterprise software business, and that this changes
  the success factors for companies. Big enterprise software companies are
  suffering, and most of the large software industry revenues are coming
  from maintenance. Sam covers several major trends in software development
  and in software business models and provides concrete suggestions for what
  software startup companies should do differently. He lists interesting
  investment areas for software tools, including: developer productivity,
  quality and security, application management, and process improvement.

John Mashley: History of Software Development and Opportunities

• Each major new computer generation:
  • physically smaller
  • less expensive
  • …
  • networked
• Each major new software generation:
  • physically bigger (code size)
  • less epxensive cycles -> use more
  • larger numbers of apps and programmers
  • adds radically diferent apps (successful things don’t go away)
  • adds new markets, problems, and tools
  • networked -> more distributed dev
• History of software dev
  • batch jobs
  • share mini
  • share lan
  • own pc
  • own clients
  • share WAN
• constants, cycles, changes
  • small is beautiful 1977 -> 2004 deja vu (agile software dev)
    • we haven’t improved human management issues much
  • programmer producitvity: 10:1 or more differences between people
    • sometimes finding really good programmers is the best single thing you can
      do. that’s an impetus of distributing development across the world is to take advantage
      of really fine programmers, and differences are huge.
    • some people are 100:1 better, write 3000 lines of code in a few hours and
      have it right first time.
  • requirements always hard to get right
    • error cost and impacts escalate
  • problems -> tools -> enable bigger problems … and cycle
    • some things a handful of people build today we couldn’t have done with
      100s 30 years ago
  • one person -> several -> many contributing to a software project
    • at Bell Labs, a memo said this project plans on 300 people working on
      something, and in my experience only 250 people can work together
  • same room -> same building -> … -> worldwide

Following are several talks with different aspects of problems and solutions for
distributed development and investing in software.

Going first: David Hartford is an old software guy, then lawyer, vc, now back home
as CEO of a software company.

David Hartford: Detecting Requirements Errors

• Detecting requirements errors

  • $Billions spent on tools to support millions of developers every year
  • $100B flushed every year in US alone. Not fault of programmers or
    developers. But the requirements they are given aren’t right.

• Do your business people and IT people understand each other?

  • Answer to this is usually "no".
  • Dilbert cartoon: Pointy haired boss says "we’re outsourcing half of our programming to elbonia to
    take advantage of the time diference." Elbonians says: "I have no idea what they
    want. Let’s pretend we died."
  • With distance, it gets increasingly difficult to fill in the gaps.
  • The problem: Missing, ambiguous and contradictory requirements.

• Where does the money go?

  • 60-80% of project failures can be attributed to requirements errors.
    at the very front of the process. not the arch, dev, or testers!
  • The biz people say you didn’t understand, the IT guys say you didn’t
    explain.
  • Scope creep and rework usually mean you didn’t find the errors it the beginning
  • Good requirements are hard.
    • requirements definition and validation cycle: informal primarily text
      requirements -> spec -> formal spec documents ->
      validation -> visual inspection and review of spec for errors -> elicitation
    • This is a laborious process, with error detection that’s bad.

• State of the art in requirements is Use Cases.

  • Engineering Change Order use case example.
  • Do you see the errors?
  • Given time, you might find some of the errors.
    Many of them you don’t find until 6 months later.
  • What if you could turn that text into a visualization of what it really
    communicates? That’s what N8 software does, it takes the text, structure of
    what was said, turns it into a mathematical model and gives a visualization
    of what was there.
  • Breaks, actors that don’t do much, inconsistencies, improper requirements language.
    • If you develop something from this, you won’t be successful.
    • But if you can take it iteratively, fix it, through it back to your user,
      you can do elicitation easier. Work with english, and throw pictures, not
      words, back at your stakeholders.
    • After 30 mins with stakeholders, you fix the text, get real diagram showing
      what they want.
    • We use this methodology writing our own software.

• N8 is getting platinum customers so far.

  • We’re just starting to look at outsourcing.
  • Is there a way to get your stakeholder to stay involved in the process.
  • Maybe that wouldn’t be such crazy talk anymore.
  • "mind the gap" between business and IT.

• Live N8 Scenario example: take use case written in a book.

  • First thing our software does it figure out actor, object, function, modifier.
    who is doing what to whom, in what sequence and under what conditions.
  • Also get a use case diagram (good for estimation).
  • Activity diagram: users, swimlanes across the top. System is on the right
    hand side.
    See flow of activity from applicant, and see flow of info across the swimlanes.
  • When you shrink it down, you can’t see words anymore. But you can see
    issues. We’re looking for happy days scenario: start at one end, get all the way
    through.
  • First thing we find out is something is passed to the system, and someone
    else does something totally disconnected. We teach our users they want
    to find out who is actually doing things.
  • Using "presentation view", we see customer signs loan agreement. In the
    rest of this, the person that starts out is called "applicant". Ask experts in
    the room, is applicant the same as customer, or are they different people?
    They say ah, we’re using different terms.
  • So we search and replace the term, and then remodel.
    Now we see one less swim lane, as everyone is stacked up under applicant.
    But we have a break. If the loan officer approves the loan agreement .. once
    the applicant finally signs the loan agreement.
    But missing info: how did the applicant get the loan agreement.
    So we insert sentence: "the system then sends the loan agreement to the applicant".
    And remodel to find out if that really works.
    Now we have a happy days scenario.
    (Sam: that’s cool!)
  • Once you get some requirements done, this is what the loan process looks like
    (shows complete loan business proc created using N8 Scenario, very cool)

• Think about being in a room with stickies all over the place. 3 hours to 3
  weeks later you say "oops" have to ask them about this.
  Now you can do this in a jam session. Or save to pdf and send it around.
  And instead of just connecting the dots, you can accelerate the elicitation
  of the real requirements.

• Case study: a real story, a BPR activity. We had 1 computer system, 4 people. It had a
  happy days scenario. But we then took the diagram and text and sent it to
  other stakeholders and ask is that what’s really happening?
  1 week later we had 6 computer systems and 9 people inside and outside the
  company!
  Think about what this means about the scope of the system!!

• Demo of N8 on VC process

  David types in the following, a bit at a time. N8 Scenario shows the model
  after each sentence.

  • Version 1 (the real happy day scenario)
    • The entrepreneur gives a presentation to the VC.
    • The VC likes the presentation and gives a check to the entrepreneur.
    • The entrepeneur then gives its stock to the VC.
    • The VC then sells the stock in an IPO.
    • The VC gets lots of money from the public market.
  • This elicits feedback, more steps.
  • David show smuch longer more complex process, developed using N8 Scenario.
    (we took this from a website of a particular VC firm)

• Conclusion:

  • We’re getting requirements in the way they want to express them, implicit,
    informal representations, and converting it into explicit, formal
    representations.
  • We think this will be great for the guys not just next door, but across the
    world.

• q: Does it take care of the grammar?

  • It doesn’t check your grammar as you have a grammar checker behind it.
    But it will show you how it is being communicated to others. It will reflect your
    own ambiguity back to you so you can fix it.

Steve Mezak, New software tools for outsourcing your software development

Had gluttony for punishment of doing 7 startups.
Author of upcoming book: risk-free outsourcing: how to use global talent to
create great software

My company helps companies with outsourcing.

• the promise of outsourcing
  • what if you could:
    • have a team of the best eng in the world?
    • direct their work to get reliable on-time results?
    • hire them at a third the cost you are paying now?
  • That’s the holy grail. From my experience, it is not a choice, but you had to make
    it work. Tonight I’ll talk about tools that are out there to help with the
    process.
• Outsourcing challenges

The boss and/or board says: thou shalt outsource! now how to do it?

7 deadly dangers of outsourcing software development (paper available from steve)

1. ignoring outsourcing
2. hiring the wrong team
3. being promiscuous with your IP
4. not knowing what your software should do
5. meager engingeering management
6. mediocre methodology

• #3: protecting your IP
  • fortressware (stealth software co): system allows you to prtect source code IP and
    data from internal compromise. increases control and visbility of remote team activities.
    (doesn’t leave that machine)
  • palamida: detects, manages, and reports on the third party, commercial and
    open source components that may exist in your software’s code base
  • black duck: software compliance management solutions.
    govern how software assets are created, managed and licensed.
    they maintain a database always updated so you can constantly evaluate your
    source code.
• #4 create clear requirements
  • N8 Scenario: would be an excellent product for this.
  • TRUEreq: web-based issues and requirements mgmt tool for PLM
  • iRise: lets you sketch the screens of your software, link them together so you can
    simulate the actiivty of what the software is supposed to do.
    • get biz people online and get their inputs.
    • studio product.
    • simulation platform.
  • macromedia: use dreamweaver to create and link HTML mock-ups of your screens
    • demo functionality and illustrate "use cases" and "user stories"
    • can get templates to make it look decent, add buttons and text
    • fantastic for doing demos to your VC before you have software running
    • when you’ve gone through those iterations, then you can use screen shots to
      illustrate your requirements for software or other artifacts you want to create.
  • Breeze: does video and voice over web so you can do collab in real-time across internet.
  • Another approach, many startups and CEOs can’t write anything down.
    For one CEO, I asked the Use Cases, he said it’s too big for that.
    he found a development team in US, did vulcan mind meld to transmit his requirements. they
    came up with a prototype, but that’s inefficient and unworkable process if
    you’re going to use outsourcing.
• #5: Effective engineering management
  • Artifact ($500/month): pay as you go, on demand service. requirements mgmt, doc mgmt, issue
    tracking. low cost way to do this mgmt.
  • role-based dashboards and project mgmt tools.
    normally teams cobble together open-source tools, cvs, wikis.
    Some people said "gotomeeting", basecamp, cvsdude, skype, and yahoo doodle
    give us a rough and ready collab platform for under $100 a month.
  • collabnet: put those tools in a hosted environment you can use
  • datainfocom: more BPO moving offshore.
    • mathematical models that monitor and predict when things are going to go wrong.
    • identifies root causes of the problems using proprietary algs
    • recommends optimum corrective actions
    • working with Dell and GE.
• #6: use agile-aware methods
  • agile development methodology
    • takes your dev, splits into short iterations.
    • take list of features, how many can we get done in the next 3 weeks, then work only on
      those high priority features and functions in the next 3 weeks.
    • involve the biz people in the development process directly.
    • a little debate as one of the principles is eng and biz people working
      closely together. how can you do that with offshore outsourcing?
      we need to find a way as there is too much of a cost advantage of
      offshore. webex type tools and breeze will help with that.
  • optimalJ: accelerates j2ee dev
    • generates apps from visual models
    • implements model-driven architecture (MDA)
  • rally dev: on-demand, software lifecycle management. accelerates agile software development.
  • versionOne
• conclusion
  • accelerance: outsourcing advice and services
  • full spectrum of outsourcing strategies
  • 17 proven teams in 14 countries around the globe.
  • Practical (and free) info about outsourcing from accelerance
  • writing a book
  • Steve’s blog discusses the above tools in more detail.

Sam Jadallah: Investing in Software

The internet has changed everything

I was an eng and dev, ended up at msft in 1987. It was exciting to see birth
of that industry, watch how things have changed. There were inflection
points along the way: the pc, client server, and now the internet.

The internet has brought profound changes, both to software industry.

• summary
  • enterprise software is broken
    • many challenges have come together to create mature, slow growth industry
      dominated by a few companies.
    • msft announced in earnings report today that in entire IT industry, Microsoft
      garnered 24% of all the profits! so not all that healthy an industry…
  • new business models drove changes
    • open source
      • impacts investing strategy. not necessarily an investible theme
    • SAAS, advertising are key forces
      • changing the way enterprises consume software
      • also, 3rd party payer model gives users powerful service for free
  • distribution channels are completely different
    • in the past it was standardized. resellers, 2-3 tiered channels. SI’s did
      a lot of the integration work.
    • now, the channels have flattened.
    • E.g. for findbugs software, sourceforge is now the major distribution channel.
      Two guys working on a project now have the same distribution as a large business.
  • web services matters.
    • vision of object oriented, capabilities between people who don’t know each
      other, is now here
    • rapid intergration
    • lightweight toolkits
    • transformation of SAP and Amazon to Platform services
      by tapping into existing legacy infrastructure that you have
    • someone wrote an app (delicio.us library) to catalog all the books in his house.
      bar code scan the book, do a lookup to amazon api, then just stick into his
      catalog. powerful stuff happens in a way nobody would have predicted.
  • customer controls the sales cycle
    • completely flipped the buying cycle. you can’t push stuff to customers
      anymore, they simply buy at their own pace, way and method. if they don’t want to buy, they won’t.
    • they were burnt through the bubble, Y2K panic cycles, and internet
      cycle. Now they’re in charge and we have to recognize this.
• Maturing software industry
  • Software is now the maintenance industry.
    • Largest percentage of their profits come from maint, not new licenses.
  • deal size shrinks
    • when Microsoft bought siebel, I signed off a $24M order (services, software,
      everything). No longer will that happen at Microsoft. (That was my last
      mistake.)
  • 80% of sales force focused on 20% of revenues
    • So cost and effectiveness per sales rep has declined to point it is not profitable
  • customers have become integration oriented, not replacement oriented
    • A CIO says I have no new initiatives. I bought stuff over 5 years, now have
      it working together. I’m building things on top, using WS. But not buying
      anything in the core as it has to all inteegrate, if not I am not buying it.
• Only the REALLY BIG matter
  • Software industry should probably have 20-25% of the number of companies it has
    today. too many companies of all sizes.
  • Because of shift to maintenance and they don’t get growth, it’s become hard to get
    profitable.
  • In order to get a 33% margin, you have to be >$5B software company!
    • The companies between 1-5B still struggle to be interesting businesses: 18%
      margins. Below that not even profitable.
    • My friend at Morgan Stanley said his biz is going gangbusters as every
      company is trying to buy others and sell themselves to get sustainable on
      their own.
• Scale AND Margins: tough to do
  • Difficult to be of scale and maintain growth.
    This biz was exciting for many years as it was incredibly fast
    growing. people paid for growth.
    Microsoft produced enormous profits. Now its stock hasn’t moved in 5 years as not
    enough perceived growth.
  • Many scale players are challenged for growth
  • software conslidation targets have generally been subscale or challenged by growth
  • Ron Weissman: Notice that software tools vendors are in the bottom left. Borland, rational.
    Subscale margins and low growth.
  • Companies haven’t gone public lately. There has been M&A, but massive transformation in the industry.
• Old to new order of business
  • Old:
    1. Get a few large marquee customers
    2. meat-eater sales reps
    3. ave sales price > $1m
    4. replace existing software
       • what you have today is really bad, let me show you what you can do now!
       • customer sees order of magnitude improvements, rip out the old stuff
  • New:
    1. bottoms up and SMB focus
       • e.g.: N8 selling to a group, not a top-down sale where they standardize.
         that’s the way we sold in the early days of the software business.
         Microsoft originally always went aroud the CIO, who never had priority to improve
         productivity of the users. Did end run around CIOs.
    2. telesales and marketing driven
       • just can’t afford the same sort of sales and marketing as in the past.
    3. ASP < $250K (for initial orders)
       • you keep getting that with intention of getting distribution
    4. Integrate quickly or not at all
       • more interesting is the latter. come in as a point and don’t try to
         integrate…
• Software tools investing
  • Lots of areas VCs invest. this has traditionally been a tough area for VCs
    to invest.
  • development managers think they can build software on their own.
    • At Microsoft, Bill Gates would always say I can build that over a weekend, why
      are you spending $1M to buy that software?
  • Interesting areas:
    • developer productivity
      • innerworkings (mdv portfolio co): hosted, integrated training
    • quality and security
      • finding bugs, improve quality right from the start.
    • app management
      • n8 fits into that
    • process improvement
    • "tools": a tough term for VCs.
      • When I look at tools, they’re tightly integrate with a platform. Platforms
        are hard for software startups to sell. Platform vendors like to give away
        tools. So Microsoft, Sun like to give away tools to protect the platform where
        make all their money. Microsoft willing to make tools a $0B business if it
        made the OS business twice as large
• Software development trends
  • modular architectures
    • java, .net, python (emerging as interesting platform), PHP
  • distributed development is working
    • in the past VCs didn’t want to see this.
      I have a company with 2 people, on in SV, one in Dublin. Skype sits open
      all day long, just like two people sitting in room together all day long
      talking whenever they want. The tools, workflow, process, experience
      level is now working.
  • open source leverage
    • need black duck etc to avoid contamination, but it’s great to bootstrap
      we used to find a bunch of our investment money goes out the backdoor to
      pay for licenses. now we see a lot of these cos write none or small
      checks, leveraging in a very smart way. preserving capital.
  • services: live and constant builds
    • expertise says you have a live process.
      we have one company doing hourly builds on a live production system. sounds
      kind of nuts (I think it was nuts) but they got it to work. sign of what
      we’re going to see.
• Example of interesting new apps we see in Venture world
  • RedFin: startup in Seattle
    • pulled together interesting dynamic HTML, rich client. Can grab
      satellite view, find houses. Integrate to MLS, Google Maps,
      advertising network through diferent APIs. Tie together in a quick and
      rich way. This was their demo. But also the application, which wasn’t
      that far off. In 2 months they will have the system to a scalable
      way. They can build rich and deep apps very quickly because of web
      services infrastructure.
• Software models
  • From software as a product, to:
    • software product downloads
    • software appliance
      • eliminate the cost of config, hassle of installation support
    • software as a service (SaaS): hosted way, buyer can consume it quickly
• Implications
  • software economics have changed
    • for better and worse.
    • important to grasp those changes in startups.
  • product must be great
    • in an era where sales people can’t sell, your product must sell itself
    • 15 minute test: easy to evaluate and use appliance-like
    • live for feedback/upsell/upgrade
  • sales and marketing costs must be managed
    • can’t afford to spend the way you have in the past
    • open source models are promising but unproven
  • customers control the sales cycle
    • learn to make the best of it
    • fire expensive sales reps
    • if you are an expensive sales rep, you should be changing jobs…
  • we are at the dawn of a new era of software industry
    • embrace it, it will be exciting
    • It will leverage all the investments of the last few years.
    • radically different, and far more exciting.
    • that’s why I decided to stay in this business.

questions

1. Talking about integration, do you find greater interest in providing
   tools like business process engineering (BPEL)?
   integrating various systems, orcl back end, siebel front end.
   help user improve process flows. beyond data integration?

sam: all the CIOs are saying that’s where their focus is today. they made
investments, now focused on building customer facing apps that tie together
the various IT investments they made.
Its’ still very hard, and while N8 and others will do well it is still a hard and
long process for them to do this right.
The CIO who said he has no new initiatives then told me they were building a
new customer application that could only be made because of the previous
investments and he can tie them together.

david: I concur. People are trying to figure out what they bought. A lot of
functionality still isn’t being used. And tying into initiatives with SOX,
to see how these can be more properly managed in the compliance process.

John: At AT&T, my wife was in the department called "common language".
she found 400 systems in all the different operating companies, needed to
build systems that did nothing but integrate and talk with other systems.

2. It seems there’s a shift to more business innovation.
   Open Source is a technical capability but ultimately relates to the business
   model. So how will business innovation be a major thrust?

sam: I think the tech innovation is actually increasing because of this
whole infrastructure. Just gotten easier to do many things.
In the past we had disruptions that flushed all the old investments. New
apps did half of what the old mainframe apps did. Just trying to catch up.
But now we have all these incredible investments to leverage. I believe we
still have a lot of technical innovation happening. And it’s software.

There is a lot of innovation that’s not interesting. how much better does
your spreadsheet, word processor, email have to be for you to use it?
I’d argue a lot better, as I’m just overwhelmed.

The other part is tech innovation in new platforms like this TREO. This costs more
than the last 2 PCs we bought, and is just as powerful as PCs were a couple
years ago, and really hard to program. These sell at 4X the rates. 150M
PCs, vs. 600M cellphones sold over the last year.

Layered on top of that innovation we have biz models challenging the
industry. the biggest thing we need to do is things that matter to the
customers. many biz plans are still incremental feature improvements, and
that won’t make the difference to somebody that will buy them.

3. I agree integration is a big thing, WS and XML driving a lot of this.
   But other trends, like VoIP. And wireless phone was part of cellular
   networks. These 3 technologies are growing in parallel and are causing a change in the biz process,
   which may be different from pure simple integration. Do you agree?

sam: I think we’ve become global. It is just as easy to work distant as
with people in the next office. VoIP, Skype, IM collab. Literally I have
told my kids to go to bed using IM in the next room (It’s just wrong, I
know!). I’ve done the same on a trip to London. (PIR means "parent in
room"). These have changed our workflow. We need to embrace that.

steve: Responding to the previous quetion, offshore outsourcing is another
business model innovation. Creating a product for $50K that you can sell is
a boon for a company. The new browser apps are also great. Suddenly we’re
calling it SaaS, we’ve been doing it for 10 years but it is now becoming
widely accepted as the way to go.

sam: outsourcing not just the lowest cost labor, but to find the best
talent. keep an employee on a boat in bermuda.
If you call Jet Blue to make a travel reservation, you’re speaking to a
housewife on DSL in Utah or anywhere, routing using VoIP PBX.
Tap into a workforce, bring the job to the employee, who can make decision
about where they want to live. Reach talent pools anywhere.

john: there easily factors of 10X or more productivity difference in
software development. So if you’re a below avg programmer you better find
some differentiation. If you’re a superstar programmer, you can live
wherever you want.

ron weissman: I think things are worse than Sam says they are.
To David, Steve: what are you doing differently to counteract the reality
Sam talks about for 90% of the venture-backed, IT focused companies?

david: we’re going after department level, vs crack a 7-figure sale. And the
problems we’re targeting are recognized at a department level. We can do this with
telesales, web sales, and then move to a quasi-traditional model for
enterprise. Our product also supports 2-legged vs 4-legged sales call, as
it’s so simple we don’t require a technical sales engineer (SE) along with the sales guy.
It’s also an incremental sale. We get a sale in Q1, Q2, and Q3. Not waiting for the
big one to hit, we can get them to see the value, and go back in where it
makes sense to make the onsite call, otherwise we just call them up.
As for viral spread, we’ve gotten 5 of our customers in a viral nature. 2 from
consultants using our software doing a project and their customer asked for how
they did that. We got 2 customers when people left to new companies and
brought us in. More recently, a manager went to a new company and said:
"we’re going to standardize on this as we saved $100K in the previous customer."

We are trying to take our most vociferous champions and get them to move to
a new company to bring us in

steve: offshore companies find it hard to market to US companies. and US
companies struggle to provide value.

4. I was at Bell Labs long ago. Lately I’m doing bizdevelopment for
   offshoring. Finding that the level of dissatisfaction of companies with some
   offshore experience is high, like 35% of companies I’m talking to. Some of
   our clients used other services and are now talking to us. What’s your
   experience of customer dissatisfaction?

david: In our (limited) experience, everyone we’ve talked to who did
outsourcing, for development or BPO, had some level of dissatisfaction. One customer
said I’ve decreaesed my cost of failure, not increased my odds of success.
People tend to throw problems to someone else. You can’t throw an undefined
process at someone else, especially lacking cultural context you have here,
and hope it’s going to work well. Hopefully this is a learning process, so
the next round we’ll know what we’re doing. You always have to have the
company store before you put the franchise out.

steve: you can take advantage of outsourcing team to avoid the dangers. Like
my 7 deadly sins. In doing outsourcing, you can now adopt a process.

sam: biggest problem with outsourcing it is got hyped very quickly. I
started outsourcing in Microsoft in 1992, for tech support. An indian employee
left, set up operation to do it. Our expectations were very low, and it
worked. We started doing more and more. People got into a big process
change, didn’t know how to measure and commnicate it, overestimated the
return they would get. It makes sense, but not in every case, and we
haven’t gottent to a point to define that and that’s part of the
dissatisfaction.

John: Something comes out, good idea, gets overhyped. works well for some
folks, not others, and you sort it out. there are plenty of bigger
companies who established outsource development teams a long time ago, sorted
through that, it works pretty well.
with smaller cos, interesting thing about Steve’s efforts, it is often
pretty random. sometimes great, sometimes atrocious.

steve: I outsourced from sunnyvale to palo alto and still struggled.

sam: I outsource all the time. My associate and I prototype stuff, then he
goes to rent-a-coder, puts up a job. the most we ever paid was $350 for a
job. people work for 2 weeks. One guy fell behidn schedule, he offered $50
more and the guy had it done the next day.
Just describe in a paragraph or two and it works.
It’s so cheap we can submit it to 3 peopel and hope one works. But there is
mgmt overhead you have to deal with.

5. The big players are going after small and midsize customers. does that
   work with these biz dynamics you outlined?

sam: the big challenge is the cost of integration is really hard. hard for
msft to mange 40K developers and ship an integrated product suite.
GE is a conglomerate, don’t need to coordinate. when you get into software, if
you’re selling to the same customer base they expect it all to work
together.
that’s a really complex problem at scale of development we’ve never seen.
I don’t know how Microsoft builds the next big product that works with the last
big product. How does xbox connect with my pc, phone, server. Just
incredibly hard to do, different internal formats, optimized around
different things.

Oracle buys Peoplesoft, what do they do? Keep the codebase or destroy it?
Just before, Peoplesoft bought JD Edwards.
Microsoft has 4 such systems, which they are now trying to integrate.

Scale breaks. Economically they have to integrate. but for having a product
strategy, we haven’t hit the problem and seen the solution yet.

6. Do you really believe the slide that shows only the 5B companies will be
   profitable, if the value will be cut 10x?

sam: the big companies will survive for a long time. The cashflows on ORCL
and Microsoft are huge. THe question is how do they get growth?
the only way they know how to do it right now is by acquisition. When you’re
a 10B, 40B business like Microsoft, they grew larger then all their competitors
combined as a business. I don’t know how they get growth. Spin out
businesses. Conslidate.
But their cashflows will allow them to survive for a long time.

6. With commodity hw and systems level platforms, now we can grow downmarket
   and new geographies. do you agree?

sam: growth from a pc standpoint: 150M pcs sold. China is the #2 or #3 PC
consumer in the world. interesting market, but real challenges how to price
in these markets to prevent piracy if you’re a software company.
if we can get the price of a pc down to $100, things change, more people can
afford them.
these $600 cellphones with $100/month services are going the wrong way. In
India, I met with Nokia. they’re designing a phone you can buy for $10, be
profitable on $8/month rent. That $8 phone is far more powerful than my
TREO, with video etc. Why doesn’t that phone come into our market and take
our busienss down even further?
I don’t know how to get growth other than lowering prices so more and more
people use it.

7. If you’re saying we need another quantum price drop, then we need a
   similar drop with software price…

sam: yes, I think we’re in need of another overhaul in software industyr. it’s
stressing out. that’s why you’re seeing advertising model emerge.
I talked to CEO of GM, trying to get media inside the car. He said: "at $250/month of
ads or services, I can actually give you a car". That’s believable!
These devices, phones, cars, are just vehicles for delivering advertising.

john: Microsoft always thought the hardware should be free. Intel always thought the software
should be free…

8. barney pell: Doesn’t the belief that software must come down in price reflect a belief
   that (a) there is no more major value attainable by software, or (b) the providers
   can’t capture a good share of that value?

sam: It is true that if you can truly generate value then you can make
money. The challenge, as investors, is to show you have a must-have for that
customer to be competitive, therefore they can attract value. customers
are willing to pay for stuff that grows and improves their business.
many companies are struggling to show this.
It is tough to integrate to let them realize that value.
Many companies come in with a TCO or ROI pitch, e.g. saving a customer $2M per year.
They say all these people are doing this, we chop the process and eliminate
guys.
You walk into the customer, give them the list of the 40 names of employees
they can fire because they bought your software.
It’s mythical, there aren’t the 40 employees you can eliminate.

9. We tried outsourcing, first in ireland, then india. now in malaysia we
   own our own development team, use our own software and sell that to customers.
   some of the things like skype and glbal platforms, we see lots of
   opportunity there. we have strong global demand.
   so stop looking at US markets, turn on its side, and go for a slither and
   use global resources and people and selling.

steve: One startup I’m advising has a line in a specification saying "yes, we’ll
support internationalization". I’m saying do that quickly. Ebay lost out in
international markets as they didn’t address it quickly enough.