Varud

This is kind of funny.  I was quoted on page 12 of a slideshow produced by the VC group North Venture Partners.  I think what’s most amazing to me is that people still have time to write all of this stuff.  I’m thankful somebody is doing it and I hope that somebody also follows my call for a Walmart approach to VC.

Being that it’s a Web 2.0 world, I would be more inclined to see something along the lines of Prosper.com but instead of loans and lenders, this would be for shares and investors.  The only truly necessary thing to do on a custom-basis would be to have audited books.  Everything else could be automated including the legal forms, escrow, etc…  For the most part, Prosper.com has already done much of the heavy lifting.

The company that does this would make money by charging a percentage of the investment.  Because investments would be done on a bidding basis, going around the system would be extremely difficult and not efficient.  Minimum bids would be really low (i.e. $100) in order to get traction.  All businesses would have to have a C Corp set up – the site could even force that C Corp to be a Delaware corp for consistency.  The thing I’ve noticed is that people just want to be told what to do – they don’t want alot of custom choices.

The one choice that might need to exist is whether the investment should be East Coast style or West Coast.  That option would probably introduce enough complexity for the VC groups to justify charging their investors for their services

Ok, I don’t normally go into politics here, but I think this is important.  Thankfully, Governor Paterson has declined a seat in the Senate for Caroline Kennedy.  She was always a lesser choice and I’m glad he held firm in this conviction.  She would not have been the best person for this state, nor for this country.  We have many gifted politicians from which to choose – she is not one of them.

I think there are two excellent options.  First up is Carolyn Maloney from Manhattan.  She is the first woman elected to her district and generally holds good middle class opinions.  However, she has been holding on to a safe seat for twenty years now and I wonder if a downstater is the correct choice.

Kristen Gillibrand was also the first woman elected from her district.  She graduated Magna Cum Laude from Dartmouth and is an upstater.  She fits all the criteria for a solid choice by Paterson.

I just hope he doesn’t choose Cuomo.  Attorney General Cuomo hasn’t really accomplished anything as far as I’m aware – he’s simply the son of a famous governor who has played all his cards right.  He essentially followed in the footsteps of Spitzer with his anti-capitalist/Wall Street vitriol and kowtowing to the power brokers of New York state.  We can do better than that.  We did better than that by booting the disgraced Spitzer.

So far, Paterson has been my favorite governor ever – I have a feeling he’ll continue to impress.

I noticed today that Meetup.com is down.  I’ve noticed this before with sites that are growing but aren’t yet “big”.  Twitter and their fail whale is of course an obvious example.  What do these sites have in common?  They run MySQL.  I’m a big fan of MySQL and have been using it since version 3.23 in 2001.  However, it simply is not a robust database when scaled horizontally unless you do alot of trickery.  The main problem is that it cannot be clustered without giving up referential integrity.

For those who are unaware, referential integrity is when you have two tables with data that are in lockstep with eachother.  These tables are just like spreadsheets in Excel – a set of columns and a set of rows.  Let’s imgine that there is one table for all the desks in a building and one table for all the employees in a building.  Each row in the ‘desks’ table would have an entry for the location of the desk and who sits at that desk.  Each row in the ‘employees’ table would have an entry for each employee – first name, last name, etc…  If a database has referential integrity between the two tables, you can never have an employee in the ‘desk’ table without that employee also existing in the ‘employees’ table.

Many people say “Good Riddance” when they give up referential integrity.  It obviously takes alot of processing power to maintain those relations and sometimes there’s a reason that an application needs flexibility in the situation.  For example, if the ‘desks’ and ‘employees’ tables are in the same database as a ‘payrolls’ table, there could be a situation where we need to be able to deal with the fact that an employee was fired by deleting them from the ‘employees’ table even though that person has now become a consultant and still uses the desk.  The biggest problem with database design is that as the application matures, business logic put into the database can become outdated and difficult to extend.

However, most modern applications accept the need for deep reliability about certain fundamental data points.  There is typically a situation where you know that you need that integrity and then you’re stuck, in the case of MySQL, between choosing that integrity and choosing uptime.

The best solution to this situation is to have multiple read-only MySQL servers and 2 write-only MySQL servers acting as a master-slave.  All typical web requests would hit the read-only databases.  Any logging type requests would be queued for insertion into the master write-only database.  There are 2 of these databases so that one can be dropped at any given time.

Cheers!  Meetup.com is back up.

A friend of mine was laid off yesterday.  He worked at a company doing enterprise search services for large organizations.  His focus was on the media industry in NY – which I guess isn’t buying entreprise search as much as other industries.

I find this a bit disconcerting.  Hopefully for the city, that company had other problems.  It sounded like they didn’t get the VC round they were hoping for and that’s certainly an issue.  I’m surprised that a company at that stage wouldn’t be cash flow positive.  They had well over 100 people.

If anybody needs work, let me know.  I run across good jobs sometimes.

I just got this email from the CEO of Crowdspring asking me to vote for them for some upcoming awards.  Here are the links:

• http://www.crowdspring.com/wired
• http://www.surveygizmo.com/s/89291/what-works-people

I really like what these guys in Chicago have done with this idea.  I voted for them and I hope they win.  Crowdspring epitomizes the intersection of crowdsourcing and expertise – which is where things are heading IMHO.

Here is the link to their post:

http://blog.crowdspring.com/2008/12/05/help-us-help-you/

I’ve burned way too many hours on version control over the past few months, I just wanted to write a short and sweet post to let people know what’s going on.  First off, there are three valid options for version control these days:

• Git – Used by the Linux kernel at this point.  A super-flexible system for distributed development.  The one drawback to this system is that distributed development is built right into the system so that it is very difficult to get acquainted with for the casual developer.  I do not recommend this system unlesss the user is technical or the project already exists in Git.
• Subversion (svn) – I’ve used svn for many years now.  It’s a really fantastic system.  It is used by Google Code (version 1.4.0 btw) for all of their hosted open source projects.  I highly recommend this.  Version 1.5 is probably much better but for this particular type of software, I prefer to follow rather than lead.  I’ll move to 1.5 when it’s standard on Mac and Google.  Larger groups would benefit from going to 1.5 right now because of it’s better merge handling.
• Mercurial – I didn’t play around with Mercurial much but it looks fantastic.  It’s especially good for Python developers because of the large Python/Mercurial community.  I would highly recommend this package for somebody who isn’t already working with the other two – based simply on what I’ve read and community feedback.

The second large problem is hosting.  I tried to do my own hosting but it’s a real pain in the #$@$.  At first, all is well.  Then, you have to do something weird (like delete and re-add a directory) and all of a sudden you’re out on a limb – wasting time – not getting things done.  Here are the options:

• Github – Github is hands down the best option for Git projects.  All the open source projects using Git are there and a small fee can be paid for private projects.
• Assembla -  Lots of people recommend Assembla.  I haven’t tried them but they look fine.
• SpringLoops – This is a really great svn host.  I’m using them right now.  It’s pay as you go which is a really good option (free for small projects).
• Svnhub – This is Github for svn.  When this is live, I will definitely check it out.

None of these services are more than a few dollars a month – even with many users.  Don’t waste your time hosting and also don’t waste your time with a half-assed solution from someplace like Dreamhost.  These sites are more than just version control hosts, they also give administrative control, good workflow, project management hooks (to Lighthouse, etc…).

Lesson learned.

The news this week in NY is focused on the budget deficit for the MTA.  The NY Times City Room has a nice article about the current thinking of the Governor, the Mayor, and the various boards involved.  I agree with this commenter that a toll on the inter-borough crossings (Brooklyn, Manhattan, and Williamsburg bridges) is a bad tax.  This mainly lies around the issues that tolls are inefficient.  40% of the income would go to maintaining the toll.  That doesn’t include the increased waste caused by sitting in traffic, the BQE backing up, and Williamsburg being full of cars waiting to pay the toll.

The best solution is a gas tax.  Increasing the gas tax would cost virtually nothing and it would raise money in a much more equitable way.  The tax would ideally come from all the counties where the MTA operates.  If that is not possible, the 5 boroughs should be where the tax is raised.

The underlying problem of inefficiency in spending is still to be addressed, but honestly, I’m not sure where this waste is.  MTA workers may get paid slightly more than they should, but it’s not over the top.  Every year, the Post uncovers some $1M that was wasted, but that’s the nature of a billion dollar operation.  That’s not waste, it’s just organizational inefficiency – it can be battled but not defeated.  The only area I see waste is in the bus system.  On 23rd street, every morning, there are about a dozen buses pulling up every 5 minutes – each with just one or two passengers.  I presume that most of the passengers were dropped off in midtown but if this is the case, the buses should terminate at midtown and the passengers should take the subway or a local bus.  Otherwise, express busses that are terminating in Manhattan should accept local passengers while in Manhattan – that would be another good option to reduce waste.

In a follow up to my previous post and others’ about the New York startup scene, Gary’s Guide has put together a fabulous central resource for technology companies.  This is a fantastic site, and even better, it’s not in Flash.

I recommend everybody to check it out.

There’s been alot of buzz about Change.gov over the past few weeks (I can’t believe it’s only been that long).  Lawrence Lessig is excited about the use of the Creative Commons license (which we also use) and Tim O’Reilly really wants them to use a version control system.  I like Tim O’Reilly, and he’s made great contributions to the technology community and society at large.  However, Larry Lessig is much more measured in his responses – rightly so.

Using the CC license is definitely the right way to go on the site – it makes it clear to people that the content on the site is public and that there is no hidden value to the information there – that one day a user couldn’t be sued.  It also allows for instant and automatic use of the content (i.e. full redistribution by bots of the content to users behind government firewalls).

In addition to issues of applicability, the use of the CC license is also a signal that Obama wants the entire government to move towards an open by default stance.  This movement was started by Carter and others with the Freedom of Information Act – these are great contributions to a free and open society.

Version control is a separate issue – no less important.  Unfortunately, I think O’Reilly goes too far.  I don’t disagree with anything he says, but his focus is not tight enough.  Just as Lessig acknowledges that the CC license won’t apply to otherwise copyrighted information coming from other departments (yet), the same measured approach needs to be taken with versioning.

O’Reilly wants everything to be shown as a versioned document, specifically legislation.  While that is great in theory, it doesn’t even meet basic tests.  Most legislative disputes are discussed either by informal discussion or in private committee.  The deliberative process needs to be kept secret in order for legislators to horsetrade, bluff, bandwagon, and balance against and with eachother.  This may sound Machiavellian, but it’s important.

Really, there are two applications for revision control.  The first is to document why and how a document changed.  This is what I refer to above and I think is not realistic to implement it – regardless, it confuses the issue.  The second application is to track official and approved changes to a standing document.  This is very much how a proper blog is maintained – or in an old fashioned sense, how newspapers display retractions.  O’Reilly and others need to push for version control on standing documents – where the need is great and the drawbacks are few.

Having solid version control on these standing documents makes changes more transparent and provides accountability to the people.

One of the first things that even an ultra-light startup needs to worry about is a checking account.  I already had an account with Citibank, so I figured I should go with them.  That way I could manage my personal and business money easily.

Citibank does not work that way.  In order to access business funds, one has to log into the business interface.  This thing must have been designed in 1998 and hasn’t been touched since.  In addition to a terrible user interface, it loses my cookie every few days so I have to put in a whole slew of information to get back on the site.  They give me a random number generator to get onto the site as well as me having a password – that should be enough for security sake.  Google is smart about these types of things and they actually design systems (like Analytics) for different size organizations.  If I had millions of dollars in my account, maybe the extra hurdles would be worth it.  However, with about $3k in the bank, what’s the point?

Another issue is the $20 monthly fee.  Chase is much better with regards to these issues and I would recommend to people that they consider Chase or BoA before Citibank.  Local banks may also be a good option but I don’t know.