Saturday, June 15, 2013

A little tech interlude

‘Just think of PRISM as your fully automatic, free, government sponsored cloud backup system.

prismThe Guardian continues to publish information about the scope of PRISM wiretapping, it seems clear that they are caching vast quantities of data for retrospective mining.  The purpose of the Bluffdale Data Center seems suddenly obvious;  the EU is rumbling that several sovereign constitutions may have been breached. 

A TWIT interview suggests that the NSA is, in all likelihood, spitting data off the fiber optic backbones just upstream from Google, Facebook, Yahoo, and others, and that would be consistent with known taps on undersea cables and satellite ground stations.

I expect that most people will eventually settle into the position of allowing anything that makes them feel safe the government echoes that I you have nothing to hide, you have nothing to fear.

<sigh> Just like in Turkey.


Is it possible to create a startup without basing it on a supposedly ‘disruptive’ technology?

Geek techI find myself shifting tech allegiance these days.  It’s not planned, but I like keeping up with cutting edge innovation and in-depth information. 

For years, computer innovation was Microsoft, just as it was IBM a generation ago.   I used their tools, learned their languages, configured their systems, and kept up with each new release.  But the change cycle is getting longer, the improvements less compelling.  Google seems more creative, aggressive, and pervasive: if I want to be part of it, it’s time to learn Java and Android.

Similarly, TWiT was my go-to source for weekly podcasts discussing technology and society.  I’ve started following more programs from alternative 5by5, more as a supplement than a substitute.  The perspective is younger (sometimes naively so) but the news feels deeper and more current.  There are alternatives, TPN and others, that I’m still exploring.


Gnod is a search-engine to find things you don't know about by asking what you like and what you don't like.

Similarity SearchSimilarity searches are based on finding data that most closely matches a given set of examples.  Nearest neighbor techniques which sort candidates based on the number of matching qualities are one example, but there are many others.

These are applied to problems like Find Music / Books / Movies that I would like”.  Netflix famously sponsored a contest to predict movies that people would like, and most tech companies are rolling out music services based on individual and social preferences.

In other domains similarity searches don’t work nearly so well.  Looking for news stories like “the ones you prefer” leads to an recursive chamber of single viewpoints and biased facts.  Media programming designed to appeal to the broadest possible audience ends up satisfying nobody.

I’ve been trying to figure out the difference: what problems are candidates for similarity searches and which ones aren’t.  Not much progress yet, but every time Facebook serves me a friend, or Google an ad, I know that nobody else has solved it either.

Friday, June 14, 2013

Tom Swifties (Nederlands)

Tom SwiftWhile waiting to see my huisarts this morning (VGZ required a renewal of my physiotherapie prescription for my ankle) I picked u the office copy of Groet Wiek.  It’s a buurtblad, one of those little magazines, half adverts, half little lifestyle pieces, that pile up next to chairs in waiting rooms and restaurants.  This  week’s had pages on building a droombadkamer, the perfect haven after een lange, drukke dag, a Gehaktballetjes recipe (met Maatrichterstroop, if such  thing even exists), and society photos from a recent Groot Wyck Borrel.

So, I’m idly reading through this, skimming and dipping,when it occurs to me that I am idly reading through this.  No dictionary, no subcaptions, just ordinary, everyday reading.  ‘granted, it’s a buurtblad, but Fantastic!


I had always expected that when the day came that reading was effortless, then my Dutch language learning would really accelerate.  Vocabulary would build by association; grammar through repetition.  There would  be news to discuss, references to share in conversation.  Context and colloquialisms would color my world.

But, damn it, I’m not finding that facility with reading and listening is helping my writing and speaking as much as I hoped.  Finding the words, composing a sentence, conjugating and adding –e appropriately, running the pronunciation, then saying something, is still a enormous task.  So, too often, I fall to using shortcuts.  I re-use a few familiar words (gaan, in particular) because I know the tenses and constructions associated with it.

So, I‘ve embarked on a Tom Swiftie project.

Tom SwiftyTom Swift was a boy genius in 50’s literature, famous for not being able to simply “say” anything.    The author refused to use the word ‘said’, so Tom ‘exclaimed’, ‘cried’, ‘joked’, ‘expanded’ in endless variety.  It spawned a series of sequential puns, known as Tom Swifties:

  • "I might as well be dead," Tom croaked.
  • "We just struck oil!" Tom gushed.
  • "It's freezing," Tom muttered icily.

So, point being, I’m making conscious effort to find equivalent alternatives for my go-to words.  So, for gaan, perhaps vertrekken, lopen, or bezoeken.  For gooien, maybe werpen, smijten, slingeren.  There are useful sites to help find synonyms and antonyms in Dutch, and teasing out the shades of meaning in my van Dale is becoming a bedtime wandelen.

I’ll give myself bonus points if I can make a real Swiftie in Dutch– so far they’ve been few and far between.

And, yes, I was a huge Tom Swift Jr. fan growing up.  No surprise: science, engineering, travel, adventure?  It was the future to me.

Wednesday, June 12, 2013

Qualifying consultants

My friend Martin Rigby sketched out a  Funding Escalator for me a few years ago, illustrating the progressive process of getting money into a startup. (Martin recently published a book expanding on these ideas, Show Me The Money, with Cambridge Angel Alan Barrell).  I use the diagram in my teaching because it’s got two nice touches of reality: there are gaps in the funding that you have to be prepared for, and the lower left box Consultancy or other second income to live on.Funding escalator

Most entrepreneurs take on small consulting jobs, probably around 20% of their time.  More than that and the startup investors start to wonder about your commitment to their project; less and I can’t accomplish much.  I charge around 200 euros an hour if I am creating content or analyzing data, of half that to review, connect, or advise.

Consult 1We naturally compete with the big groups that consult  full-time. So I was intrigued to get a chance to review two bids from a friend asking which they should accept for their new laboratory analysis startup.  In 2.5 months and 23,000 gbp, the first would

  • Generate a basic regulatory roadmap to help understand the requirements for technology development from laboratory R&D stage
  • Provide our view on the current status of the technology and assay positioning including analysis of competitors with similar assays including performance data, assay formats etc.
  • Prepare a power-point style slide presentation that summarises the outputs
  • Create a product development plan based on a stakeholder-agreed strategy to develop the technology to a significant commercial milestone
  • Prepare a list of potential funding sources and collaborators to further support the development process

In 2.5 months and for 84,000 gbp, the other consultant will:

  • Review current data and documentation pack
  • Analyse assay configuration(s) to better understand product format and map likely development route
  • Perform a basic technical analysis of competitive products to understand key technical aspects
  • Generate a basic regulatory understanding, outlining key features of the route-to-market in US and EU and clinical data requirements
  • Develop a top-level product development plan that outlines key milestones

I was actually pretty horrified by both proposals.

  1. The market analysis should include analysis of the market need and patent landscape, along with the competitive analysis.  This may include interviews with customers and distributors.
  2. There is no analysis of the health care or business economics, the pricing or return on investment.
  3. There is no analysis of exit strategies, including exit points, potential multipliers, or similar deals in your space
  4. There is no plan for budgeting or fundraising, relative to project milestones, in the final report
  5. There are very few tangible deliverables: it seems like guidance that could be available elsewhere (regulatory classification and requirements), template documents (the development plans), or simple advice without any introductions.  The “list of potential funding sources” is particularly useless.

Consult 2It made me think about the purpose of hiring a consultant in the first place.  Big companies bring in McKinsey because they provide objectivity, gravitas, and gloss to CEO strategies. But for a startup, a consultant should:

  • Validate the market opportunity and timing.
  • Verify the existing unique (or not yet existing but necessary) qualities of the product.
  • Highlight regulatory and patent risks.
  • Characterize the customer and exit opportunities.
  • Estimate the timing of market entry, adoption rate, pricing, and return on investment.

It’s what I usually expect in a contact; it’s what I would do for my clients.  It’s information that a startup needs to go to investors with a clear plan for the business (and budget), to plan milestones and fundraising goals, and to argue that the market and investment opportunities are objectively assessed.

The other eye opener, of course, is that I have got to raise my prices!

Monday, June 10, 2013

Bletchley’s heirs

I’m avidly following the Guardian’s revelations about electronic intelligence gathering, the ‘Prism’ program that is causing  embarrassment in country after country (the Dutch admitted to receiving information today).  It’s a fascinating tech story, fun to watch the shifting political and business lines, and philosophically interesting to muse on the morality of it all.

PRISM 1And I enjoy everyone making fun of the logo.  Who would even think that a top secret program needs a Bond-style logo, much less say Yeah, that sells it to this one (right).

It’s almost as good a question as why the UK intelligence center, GCHQ, has a headquarters in the stylish shape it does (left). But more on that in a moment.

I tend to be a 4th Amendment hawk, the way some folks are ‘right to bear arms’ advocates.

The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated

‘Seems clear cut.  The law recognizes two different searches within this definition.  Investigative Searches are the classic ‘Get a reason; Get a warrant’ searches for evidence, given a judge’s finding of probable cause and a police officer’s participation.  Regulatory searches are broad inspections applied indiscriminately to large groups, as in airport screening and restaurant kitchens, requiring no judicial pre-authorization an conducted by government agencies rather than law enforcement.

TIAThe Prism program likely falls under the second category, justified as a reasonable search to promote the safety and welfare of individuals and of the public.  Generally, I think that the volume of material makes any detailed inspection of my personal record unlikely, and my correlated calls and emails probably don’t trip anyone’s trigger criteria.

But there’s no protec5ion if someone did take an interest, for any reason.  They could assemble a very detailed record post hoc instantly, and I’m sure that the results could be used in a creative variety of direct and indirect ways to harass, fine, and embarrass me.  How many candidates for public office or conscientious whistleblowers will be silenced because of this information?

Coincidently, I visited Bletchley Park this weekend, home of the team of codebreakers who cracked the Enigma machine during WWII.  It’s a fascinating place, still being restored, but with working Enigma coders and BOMBE decoders, the original Colossus (one of the first programmable computers), and rich stories of how the codes were deciphered.


It’s a geek-paradise, lots of lights and relays (you can see how every 60’s television example of a computer started here), very good storytellers (our guide sounded a bit like Montgomery), and surprise characters (the three Polish mathematicians who worked out the original Enigma details).

Punch cardA key innovation, though, was a vast filing system of punch cards that contained all of the facts discovered in the decrypted messages.  These were kept in file drawers, then collated to understand who was going where and doing what so that their plans could be inferred.

Just like Prism.

A lot parallels the Los Alamos story: brilliant scientists taken to a secret location where they work all day (and do amateur theater at night), the tensions between head scientists and military officers, the unlimited funds thrown at research, the eventual triumph and final prosecution of the lead players (Turing and Oppenheimer).

There is a fascinating crystal presented to Bletchley by the GCHQ: when you look in one side, there’s an encoded message; look in the other and it refracts to plain English.


And I’m convinced that the circular design of the building is a stand-in for the dials on the BOMBE.