Google’s self-driving Car: The beginning of Google Car OS?

Google already has a large share of your attention at home, in the office, and on your mobile device when you are on the go. Sure it wants more of your time in those places, so it’s going to give Chrome OS and cloud apps, and what have you. But how about the one space where you spend an hour or two of time each day with nothing to do?

You guessed it: Your car! Google is not there yet! However to do anything in the car, Google has to make sure of two things: 1- That it is safe for you to take your eyes off the road, and 2- That Google is the dominant information system in the car (translation: Google Car OS).

Creating an autonomous driving system meets both of these objectives. It won’t be simple though. It will take years before all technical challenges are resolved and regulatory bodies would allow a Fully-Autonomous car on the road. But stakes are high, so early investment is the key. By creating the autonomous driving system years ahead of everyone else Google establishes itself as the default choice of an Information System OS for car manufacturers or after-market device makers. Recall how Microsoft’s OS monopoly has allowed it to milk the PC software market for decades in a myriad of ways. Now compare the average life expectancy of a PC with that of a Car, and you see how much more valuable of a market that would be!

P.S. The image is KITT, The computer driven car from the "Knight Rider" TV series.

Why are we surprised? Google's moving to the last mile!

While Google Buzz seems to have overshadowed most of the news about Google, in the long run the news about Google's plans to offer super-high-speed Access to Internet is what deserves most attention.
This is the latest, and by far the clearest, attempt by Google to bypass carrier's transport choke-hold, or, as I said two years ago in a related blog: reach the open seas. U.S. Carrier's are fighting tooth and claw to protect their turf, and they are losing it inch by inch.
Consider the mobile domain. For years carriers dictated the types of mobile phones users could choose from, as well as the most minute details of the hardware and software features that each phone. Carriers considered themselves in full control of the access networks (plumbing), and dreamed of going up-stream to the content! But they did a lousy job of it all by trying to control all resources, build walled gardens (instead of open oases), and pinch every penny that passed through the system.

Fast forward to 2010. Carriers are on fast track to losing their control of applications and content. iPhone's ecosystem is independent of AT&T, and Andriod's independent of Verizon. Other phones (and environments such as Nokia and Blackberry) are already there or getting there. Question of content is settled and behind us. The hottest battle-front is now the last mile (plumbing). Carrier's core business is under attack, and I doubt that they will be able to defend it in the long term (15-25 years?).

Where will the next battle front open? How about mobile payments? This is an area where carriers and banks have successfully stifled all innovation in North America. Will Google partner with PayPal, Amazon, eBay, and a smaller payment system (say Discover) to break the logjam? Let's hope so.

Reliability vs. Availability

There is huge misunderstanding about Reliability and Availability. I recently came across a multi-million dollar project where the customer was asking for a 99.9% reliability for a web-based "service". I could swear they meant Availability.

I have seen this confusion often enough that I think a bit of explanation could be useful. So in a series of postings I will address the following points:
1- The technical definition of each word in simple terms, and their mathematical calculations using real-world examples.
2- How to configure a system to meet different levels of availability.
3- When should you care about Reliability, and when should you care about availability.
4- If you are drawing-up an RFP or a contract, what are the parameters that should be specified in each case when you ask for (or commit to) a specific level of Reliability or Availability?

This blog, the first in the series, deals with the technical definition of each word, and their mathematical calculation.

• Reliability: Reliability is the likelihood that a given component or system will be functioning when needed as measured over a given period of time.
  For example let's assume that you have a system with an MTBF (Mean Time Between Failure) of 3 years, or 26,280 hours. You are interested to calculate the likelihood that this system would have no outages during any 1 year observation period. To do this use the following formula:
  

R = e**(-t/MTBF)
where
R = Reliability
e = 2.71828182845904, the base of the natural logarithm
t = the observation period
MTBF = Mean Time Between Failure of the given system

Using this formula, we find that the answer to the question is 0.7165. In other words chances of a break-down during any 1 year observation period are 1-0.7165=0.2835 (~28%).

• Availability: Availability is the percentage of times that a given system will be functioning as required. The measurements that form the basis of calculation for this percentage may be discrete (number of times an engine will start if tried 1000 times) or continuous (the number of hours in a year that a telephone switch will be operational in a given year).
  In the converged worlds of IT/Telecom, very commonly the word "Availability" is used to refer to "Steady State Availability" or up-time ratio.
  There are two formulas for availability:


• First let's consider how an "unscheduled" outage can impact the availability of a system. To do that use this formula to measure availability:
  
  A = 1 - (t_outage/T)
  where
  t_outage = Duration of unscheduled outage
  T = Agreed upon window of time for Availability Measurements
  
  

  So as an example, let's assume that your contract calls for 99.95% availability in any 1 month period of time, what would two unscheduled outages of 15 minutes each mean?
  A = 1 - [(2*15)/(30*24*60)]=> A = 99.93%. You may be in hot water!
  
  
• To estimate Availability in advance use this formula:
  
  A = MTBF/(MTBF+MTTR)
  where
  A = Availability
  MTBF = Mean Time Between Failure of the given system
  MTTR = Mean Time To Repair
  
  

  Now, let's say you are using the computer system above with an MTBF of 3 years, or 26,280, and a 4 hour MTTR. What kind of Availability can you expect?
  

  Plugging the numbers gives you A = 26,280/(26,280+4) = 99.98%

---

So, let's ask a question now. Based on the example above, if you are providing a web-based service using the system above (with 3 year MTBF and 4 hours MTTR), it should be safe to commit to 99.98% availability, correct?
No, not necessarily. The calculation above applies only to the full duration of the MTBF (3 years). What it means is that on average, many systems of this type observed over a 3 year period will have an availability of 99.98%.
When promising availability one of the important factors to consider is the "observation period". As an example if your contract calls for 99.98% availability in any given 30 day period, all you need is a single outage lasting 30 minutes, and the availability for that month drops to 99.93% (as shown in the first calculation for Availability after outage).
It may not be, however, practical to ask your customers to measure the availability of the system over a period as long as 3 years, and they may insist on a 1 month observation period anyway. The question to answer is: how can you configure the system to meet this level of availability?
The answer to this question will be the topic of the next blog in this series. In the meanwhile feel free to send me your thoughts and comments.

Tactical moves, strategic consequences: Will you survive the famine to die at the feast?

Perhaps the best measure of a good entrepreneur is not how well they can forecast the future, but how quickly and effectively, in the face of inevitable surprises, do they recognize the changed realities, and how well they adapt and respond.

For the past 6-9 months all signs have been pointing to a long and painful recession ahead. Most businesses have already responded with drastic cost-cutting measures. That is a painful but necessary step. But when large established businesses take an ax and start chopping, fledgling IP based start-ups need to use a surgical approach.

Having guided a technology start-up through the 2000 bust and 9/11 induced recession, a few observations stand out and may come in handy for those of us trying to salvage our investments through today’s hard times.

First, consider three basic ideas:

1. Small IP focused start-ups live and die on the strength of their IP. Losing the IP along with the team may turn out to be an irreversible decision. Choose carefully who stays and who goes.
2. Customers are won hard, and losing them can be very painful particularly in hard times. However, they will not risk sticking with weak suppliers. You need to avoid sending the wrong message.
3. Market will eventually rebound. If you have not planned for it, you will survive the famine only to die at the feast.

So how would you plan for this process? Consider the list below as a starting point:

1. IP Triage:
   • Begin by recognizing the company’s core IP: Technology, formula, manufacturing process, marketing knowhow, personal relationships, etc.
   • Prioritize and assign $ value to the IP in the short term and the medium term. Guess the $ if you have no other comparative or market intelligence. Don’t worry about the long term unless you are sitting on something real big, or you have the luxury of waiting for it.
   • Decide what you can afford to lose forever (e.g. patents that will lose their priority filing date), what you can reinvent, and what you MUST keep.
2. Customers / Prospects Triage: Ideally you don’t want to lose any customers (if you have any). But when customers sense weakness, they may walk anyway. Decide which customers, or prospects, you must keep, and at what cost?
   Here is the shocker: Sometimes you must lose customers to survive!
   
   • You gave away your product or service to this customer thinking you’d get yourself established in the market and get a reference (debatable approach). Is that customer living-up to its promise or is it delaying revenue-generating commitments while draining away resources?
   • Have you strayed from your core mission to pursue quick cash? In lean as well as fat times, chasing too many tactical clients means one of the two things: 1- You are not focused on the core strategy, or 2- Your core strategy has to change to reflect the new market you are chasing.
     If you are not careful you could end up with one-off tactical clients sucking more resources than the revenue they are generating.
   • Decide which one of your clients will be around long enough to pay their invoices. Unless you are in the business of “investing” in your clients, try to collect more up-front, or in multiple milestones, to at least cover your costs.
3. Resource Triage: Be realistic in your need for resources at all levels. It is easy to draw-up a list of worker-bees and middle managers to layoff. But look around the executive suite. Are you keeping all the c-level guys around because you really need them? Tales of 10 man “technology shops” with 2 engineers and 3 highly-paid C-level guys are abundant, and almost never happy-ending.
4. Communicate: Once you have it all figured out, you need to communicate.
   
   • Tell your staff what your plan is for the future of the company. They can be very demoralized after seeing their colleagues, or some of your customers, go out the door.
   • Tell the customers you want to keep how you plan to continue to serve them. Customers will learn about the changes in your company anyway. Best to hear it from you first. Go meet the important ones if you must to make sure they stay calm.
   • Tell your investors. Keeping your investors informed at all times is a good practice in general anyway. Keep in mind that not all of them are represented on the board. Especially the angle investors who came in early on may need to be reassured.
5. Remain ready for rebound: Having gone through with it, you now have the core IP, processes, and products you must keep to succeed, and the team to support it. But you are not done yet. You need to remain vigilant during the downturn so you can rebound when ready:
   
   • Maintain discipline and focus on quality: Make sure processes you chose to keep are followed. Don't fall back to garage days unless you planned it. Ad hoc loss of processes or discipline makes it harder to fly again when the market is ready.
   • Continue to invent and innovate: Smart brains do not stop working just because there is a recession out there. That applies to you and your competitors. Having focused on your core mission, make sure you maintain your hard-won advantage.
     

Given the length of this recession so far, and how much longer we all expect it to last, the last point can be the most important one differentiating the winners and losers at the end of this recession. By the time we near the end of this recession, your focused execution during the lean times will have enabled you to hand-pick acquisition targets, attract capital more easily, and most important of all, go after customers stronger than before.

WiMax: Yet another attempt by Google to reach open seas?

The recent WiMax announcement (Sprint, cable form $14.5 billion Clearwire venture) has everyone talking about Sprint and the future of WiMax. One less noticed aspect of this announcement is Google's role and the possibilities that it opens up for Google (at least later on once WiMax is well-established).
First some background. It is not secret that traditional operators continue to build barriers for open competition on the mobile networks and are trying to move up the information services chain to compete in the apps and services realm.
This has only increased the incentives for innovators who are trying to find a way out of these walled gardens, and Google is no exception. Witness Google's role in FCC auction and forcing the open access clauses.

Google is also heavily pushing to free-up the white-band portion of TV spectrum opening-up next year, and has in the past tried to get into municipal WiFi networks. This foray into WiMax is just another part of a long-term strategy.

But what is that strategy exactly? Clearly one important goal is to break free from the operator's walled gardens, but could it be that a very important side-benefit or even maybe the actual strategic goal of all these attempts is to connect Google's impressive and rapidly expanding back-haul network to the consumer in the last mile.
Google is building and buying all the fiber capacity it can to connect its data and application hosting centers at super-fast speeds. It is also building hosting facilities at unmatched pace so no matter where you are, you are only a couple of hops away from a Google data center.

With these two at hand, if Google (or someone like them) figures out the last mile problem (wideband or WiMax or something else), what would keep them from developing a few simple building-block type "telephony" apps (they already have presence and the rest), and then offering telephony services to users? This is a network built for the Hulu and you-tube generation streaming video to the TV screen. Voice and even live video (on a small screen) will be lost in all that bandwidth. That is when operators realize how easy it was for their bed-rock revenue stream -bit pipe plumbing- to disappear.
Google may not want to be a Telco, but it sure has the potential to be a Telco back-bone powering thousands of micro-telcos packaging, customizing, and selling services at neighborhood or friends and family level.


But do not hold your breath. It will take many many years for the operators to burn through all their money. In the meanwhile operators will continue to insist that because they deliver the water (bits), they can also deliver meals (apps and services) that are made with that water! In doing so they only increase the incentives and the eventual pay-off for those innovators and investors who figure out how to break through their gardens.


Digg It!

Will search engines read your mind? Now that is a new way to say "I'm feeling lucky!"

Everyone is talking about personalized search, and how search will evolve in the next 5-10 years. Here are some of the trends to monitor in the next few years:

1- Explicit Personalization: Consumers continue to provide personal information about themselves in social networks, mapping and directions services, event planning services, web-based email, and even online search portals (e.g. iGoogle). We will see more and more efforts in combining information thus available in various online settings to improve the quality of search.

2- Implicit Personalization: More effective data mining and consumer behavior categorization techniques allow service providers to better track, categorize, and even forecast user’s search requests. By analyzing a user’s past behavior the user’s demographics as well as his/her specific preferences can be determined better than ever. This data allows search providers to detect shifts in temporal variations in search patterns of small groups, then apply specific information known about each user, and arrive at a prediction of the range of searches that a user may conduct, and the data they are actually looking for.
Imagine having a set of search results delivered to you each time you visit your favorite search site, without having typed a query! Now imagine seeing what you were looking for among the top ten results 50% of the times! “Search Forecasting” (I just coined the term) may appear outlandish at first, but if best technologies are those that mimic life, then best search engine is the one that mimics a capable executive assistant. The best ones can almost read your mind!

Search Forecasting advances not only impact the web-based searches, they also greatly impact those search modes where real-estate or device limitations pose real challenges. We are of course talking mobile now. It is rather clear how deeply such advances can improve the GUI usage on the mobile device, but let us consider the impact on another oft-ignored area of search: Speech Enabled Search. Currently this area is limited to Speech Enabled Directory Assistance (SEDA) and Business Category search, provided by a few large players only. Search Forecasting allows players to improve accuracy of their systems to much higher levels than possible today, reduce the cost per search, and improve the revenue per call. Such change in economies of service opens the doors to smaller players who will be able to offer Speech Enabled Search not as a stand-alone service, but as a feature of a more focused niche offering.

In the longer term today’s directory assistance (whether paid or free) and business category search will cease to exist as stand-alone services, and will be offered as part of a bigger umbrella of services where DA (or YP/Category Search) will be only one of the many “information services” offerings.

3- Federated Search, Context Sensitive Searches, and Separation of Duties
Another area in providing highly personalized service is advances in Federated Search, and appearance of search as a natural element of any service, no matter how niche. It is often the smaller niche players such as the social networks, trip planning sites, activity planning services, or even local newspaper which have more relevant information about the user’s search context than the big search engines. By combining a niche player’s context data with the “Search Forecasting” techniques, service providers can offer more and more relevant responses for search queries, and players can improve CPM-based advertising from a border-line spam that pushes anything to a very effective and in fact often-welcome suggestion-making business.

4- Going from What to Why: Today all information services focus on the “what”. You type “what keyword” when searching online, and “what business listing or category” when calling Directory Assistance.

Users, however, act just like Hansel and Gretel.
Each search they conduct is one piece in long crumb-trail that is focused around a certain intention or activity, often involving one or more transactions. By focusing only the “what” aspect of the search, service providers lose the bigger context (“why”) and in doing so, not only do they fail in delivering a more relevant set of results for their users, they also lose a fantastic opportunity for increasing revenue opportunities in each search query.

We will start to witness intelligent information services that deliver highly context sensitive “Intent Oriented Search Services” by taking into account a user’s web-based social context and information that it offers when handling a user’s search queries.

This move from “What” to “Why” will start by niche players with explicit knowledge of user’s search context (and we have seen examples of that for a few years now), and will catch like wild-fire when Federated Search and Search Forecasting techniques evolve.

Digg it!