Planet Jeffro

I was at a conference in DC recently and stood at our table pitching attendees for 3 days.  I hustled a lot - standing out in the middle of the aisle chatting up everyone who walked by: “hey, have you heard of Groundlink?”  I had a ton of fun and must have talked to hundreds of people.

Some asked if I wanted to scan their badge.  As an exhibitor you have the option of renting a badge-scanner that basically generates a database of contact info of leads who you talked to.  We were too cheap for that.

I met a wide variety of people - government employees, business people, managers, casual travelers, etc.  All of these mini-conversations yielded interesting facts: the gov’t reimburses all travel, but sequestration is forcing agencies to cut costs, have you looked at driving congressman around, certain agencies have their own dedicated fleets.  Occasionally I came across a lead: someone who manages travel for a large company or was looking for an easy way to handle transportation logistics for events.  I got their cards and passed them on to our sales people, along with some context.

But overall, I would really like to capture these little conversations and know who I talked to. Business cards are incredibly difficult to sort through later because I forget who said what.  I scrawled notes on each card to jog my memory, but having 100+ conversations per day will make your head spin.

Enter Google Glass for conference exhibitors.

An exhibitor could rent Glass + the app for a day-rate (similar to these card readers that cost $500/day).

The exhibitors at the booth would wear Glass all day and for each person they meet, it would seamlessly record:

• A snapshot of the person you talked to along with a timestamp
• Capture audio of the conversation and possibly video (maybe prompt the user to ask for permission)
• For badge scanning, just look down at the attendee’s badge and the QR code is instantly scanned and meta data captured along with the pictures/audio/video

At the end of the conference, download all the data/video/images to a slick interface that lets you:

1. Connect via LinkedIn
2. Share leads (with conversation) with other people at your company
3. Add notes
4. Email the lead directly
5. Batch download for import into Salesforce/CRM
6. Send a mass email

Maybe something like this already exists.  It would also be easy to test and see if people are comfortable being recorded: just give an exhibitor Glass and ask them to say to each attendee they talk to “hey, I’m recording this so I’ll have more context around what we discussed, is that ok? If not, I can shut it off.”

Don’t be a glass-hole.

Thoughts?

The question is not “Is CLEAR throttling our internet speed” (that much is pretty obvious given experience and the class action lawsuit).

The question is: Is CLEAR throttling our internet but relaxing the throttle when we use Speedtest.net (which their support staff instructs customers to use in order to check the speed of the internet) so as to get a much faster rating?

Let me explain…

A few weeks ago I came across this awesome app called “PlayTube” which lets you download (cache) YouTube videos for offline viewing.  It lets me catch up on all my viral videos on the subway every morning w/o having to deal with buffering issues or commercials.  Brilliant!  It is one of the best apps I own.

When you’re downloading a YouTube video with it, it shows you the download progress in MBs.

If you know anything about CLEAR and AT&T LTE, this much you should expect:

• On WiFi (ie CLEAR), the download speed is really slow. The numbers move agonizingly slow, maybe about 100 KB/s
• On AT&T’s 4G or LTE (in my apartment it switches between them), the download speed is virtually always 1MB/s (10x).  Sometimes higher.

First, I wanted to control for YouTube’s own bandwidth speeds.  So I “calibrated” by first testing on AT&T’s network so it was obvious that YouTube speeds were never the issue.  

As an aside: I have a grandfathered AT&T unlimited data plan.  But I’ve already downloaded so much through Playtube on AT&T’s LTE network that I exceeded their 5GB soft limit and AT&T sent me a SMS saying that they (AT&T) will begin throttling my LTE speeds! They won’t charge more but basically they say that I’m in a tiny percent of over-users and this is how they accommodate everyone.  By giving us slower speeds. The real joke is that I can’t even tell AT&T is throttling because it’s still way faster than CLEAR’s wifi!

Anyway, now that we know the download speeds, here’s where the story gets interesting:

When you contact CLEAR (through phone or chat) and complain about the speed being slow, one step they always ask you to do as a final check is to run speedtest.net and tell them the result.  I’ve done this maybe 75 times already.  Often, Speedtest will indeed show a solid bandwidth speed for up and down…  yet the internet still feels slow (on desktop).

Randomly one night, I used the Speedtest iOS app to check, while Playtube was downloading a batch of videos (as it is able to do it in the background).  As expected, the Speedtest app showed solid speed results (500 KB/s - 700 KB/s) that didn’t seem in line with what the speed it felt like.

The curious thing was that when I switched back to Playtube, the download progress numbers moved noticeably faster, as if simply running Speedtest had tricked the throttling into stopping, temporarily.  Then after a few seconds, Playtube speeds dropped back to a crawl.  I repeated this 30 times to make sure it was legit and *most* of the times it had the same effect.

Internet is slow (via Playtube). Run Speedtest. Internet gets way fast for a few seconds, then slows down once again.

It appears as though CLEAR not only throttles, but that they actually DE-throttle when they sense you’re using speedtest.net, (or perhaps any bandwidth measuring service), and resume throttling when you’re on virtually any other site or service.  This is really sketchy.

Here’s a 10 minute YouTube video that basically demonstrates everything I just described.  Sorry for ranting at length.

Maybe I’m missing something, so I told a few technical friends and here’s what (the brilliant) Benzado said and suggested:

The strange thing about this is that they appear to relax throttling for ALL your traffic when you run SpeedTest.

When you download a file, the remote computer chunks it into packets and starts sending them to your computer. For each packet your computer receives, it sends an acknowledgment (ACK) packet back to the remote computer to indicate it got that one. (If it doesn’t get an ACK for a packet after a set time, it will re-send.) When another computer in between on the network starts holding the packets and releasing them at a steady rate, that’s how you get throttled.

Now, what’s interesting is that I think it would be simpler to look at the packets and simply not throttle anything used by SpeedTest. Then the throttle doesn’t have to remember whether to turn on or off for a particular traffic source, and also someone looking closely would conclude the problem is YouTube or whatever. My guess is that since SpeedTest uses a variety of sites to test against (that’s what the ping step at the start is, to find the closest site) it may be too much work to manually keep up with the list of sites SpeedTest uses.

In other words, choosing cheaper automation may be what revealed their hand here.

As far as testing throughout the day; if you just want to measure speed, you could easily set up something automated to download a file every hour or so and log the speed. Just have cron run curl and log the progress messages to a file. If you want to hit SpeedTest once an hour that’s possible, but trickier.

Also, if you could do the Clear test again on a computer or two devices (so you can see both SpeedTest and the download simultaneously) it would be a lot more convincing. I’m sure someone would argue that the iOS app download only “appears” to be going faster since in the background it was getting data but not updating the UI, and when you switch back the numbers move faster to catch up. Or something like that. The point is, iOS background behavior is weird and better to eliminate it from the test if you can.

I haven’t had the energy to re-run this experiment on the desktop but you get the point.  CLEAR is the lesser of two evils right now (we killed Time Warner Cable thinking things couldn’t get worse).  I really hate not having a choice of ISPs in NYC.

My mom wrote a great piece in the synagogue’s newsletter about trying to do “Meatless Mondays”.  I totally support this. I think it’s unlikely for many people to go 100% vegetarian but well within reason to have a diet that is maybe 50-70% vegetarian.  The majority of my meals are meatless or have meat only as part of the meal.  When we buy meat for cooking, we make sure it is not only organic and hormone free but “humanely certified” (when possible).

Personally, I’m ok with the killing of animals for food. But I’m not OK with inhumane treatment of them when they are raised and handled.  My wife and I regularly put our money literally where our mouths are and pay 2-5x for that peace of mind.  As far as I can tell, this humane treatment of animals before they are killed for food is the spirit of kashrut - not just the sharpening of the knife and method of slaughter.  But most kosher food isn’t actually raised humanely. It’s raised the same way unkosher food is - they just walk the animals down different paths when their time comes. That’s the point of this article and its placement in the synagogue’s newsletter.

Go mom!

Here’s the article text:

“Meatless Mondays”

Arlene Novich

The Green Committee: A Sub-Committee of the Social Action Committee 

You may ask, “Why would anyone want to do that?” It may be for health reasons or it may be for Jewish values of environmental stewardship or the ethical treatment of our food animals. The Johns Hopkins Bloomberg School of Public Health has teamed up with the Meatless Monday Campaign, Inc. They found that we can reduce our carbon footprint, decrease precious water usage and cut down on fossil fuel dependence by cutting out meat once a week. 

Did you know that the meat industry generates nearly one-fifth of the manmade greenhouse gas emissions that are accelerating climate change worldwide which is far more than what transportation does? Did you know that it takes 1,800 to 2,500 gallons of water to make a single pound of beef, but that soy products (tofu) produced in California require 220 gallons of water per pound? (United Nations’ Food and Agriculture Organization) Let us not forget extreme animal waste, the overuse of pesticides, the use of growth hormones and antibiotics.

There are also troubling ethical problems with eating meat, including kosher meat. The meat we enjoy has an unsavory journey to our plates. Cows and chickens spend their lives in overcrowded and unsanitary conditions until they are sent to inhumane slaughterhouses. “Just because an item on our plate is ritually kosher does not mean it won’t make us ethically queasy.” (Jewish Week, “An Ethical Seal of Approval,” 2/7/2012) As a country, we often take pride in the way our pets are treated. We cringe at stories about a dog that is abused by its owner. “We protect ‘companion animals’ like hamsters while ignoring the torture of chickens and cows and pigs,” writes NY Times food columnist Mark Bittman in his article, “Some Animals Are More Equal Than others” (NY Times, 3/15/2011).

Meatless Mondays, sharing vegetarian recipes, buying organic, free range chicken and pasture grass-fed beef may help our health, our food animals, and our environment. Someday we will have the Magen Tzedek, a long-awaited kosher and ethical certification.

I have a list of 2,000 companies in NYC and I want to grab their Foursquare venues easily.  I suppose I could just use Foursquare’s API and do a search for each name and take the first venue that comes up but here’s a cooler way.

Do a “I’m feeling lucky” search for each one:

• Include name of company
• Address
• +site:foursquare.com to limit to just a foursquare site
• &btnI means “I’m feeling lucky”

http://www.google.com/search?q=2tor Inc. 60 Chelsea Piers +site:foursquare.com&btnI

Capture the redirect that Google sends by using the setting:

$ua->max_redirect( 0 );

Then look at the header of the response:

my $firstUrl = $response->header( “Location” );

This will return something like:

https://foursquare.com/v/2u-inc/4eca786a6c251306cee623f5

We just want the last part, so we’ll add a teaspoon of regex to get the venue id:

my ($venue_id) = ( $firstUrl =~ m|/([^\/]+?)$|);

The above can be done in a lot of ways, but I used:

• [^\/] to signify “all characters that are NOT a / character”
• the “non greedy” operator “?” to match “the fewest possible characters”
• the “$” matches the “end of string or line”

Also I’m a huge fan of the “do a regex, capture a pattern, then toss that pattern into a new variable” all in one line.  Not sure what this is called (and I’m sure more elegant versions exist in Python and Ruby and Scala) but I really dig it.

Rinse and repeat 2,000 times (add a “sleep” in there so Google doesn’t immediately block my IP) and we’re golden!

Huge thanks to Perl Monks for doing 99.9% of the work here.

The first was on a CNBC story about the new NYC taxi apps. Here’s my bit at the end of the article:

But these innovators might heed the tale of Fare/Share, which launched in 2010 to some fanfare—press coverage and a spot at the prestigious “TechCrunch Disrupt” conference.

The model was to pair passengers with the same general destination so they could split cab fare. But co-founder Jeff Novich found that people don’t often turn out to really want what they say they want.

“Many New Yorkers will say they want to share a cab, but a cab is often perceived as a luxury,” he said. “People overwhelmingly want it now, and want to be alone.”

That app is out of business.

The writer, Matt Twomey, emailed Fare/Share asking how the business was going and for some stats.  I said we weren’t operating anymore and gave a little detail around why.  I guess this would be yet another “why we failed” post:

Fare/Share launched in June 2010 to some fanfare. We were at TechCrunch Disrupt, presented at the NY Tech Meetup, had a story in AMNY, WSJ and even an interview in CurrentTV.

Unfortunately, a few major issues didn’t allow us to achieve traction.  Many NYers will say they want to share a cab but a cab is often perceived as a luxury and people overwhelmingly want it now, and want to be alone.  Sharing a cab forces you to compromise on both.  

Fare/Share faced a major “network problem” where we could not match users with each other. We didn’t have nearly enough active users.  One way to tackle this would be by limiting time and/or space - the app only works at Columbia University and the Lower East Side, for example, and only on Friday nights.  Then there is a higher chance of matching.  

Ultimately, Fare/Share may have been trying to fix a problem that simply doesn’t exist.  It was filling a need people thought they had but don’t actually want to fix.

There certainly is a lot of potential.  We crunched GPS data from the TLC and found upwards of 30-40% of taxi rides started and ended within a few blocks of each other, and departed within a few minutes of each other.  

We actually licensed the app to a cab company in London and worked with them for a year. But they just never got around to launching it, so we never got to see how the London market might have done.  Combining taxi sharing with an actual service could greatly improve uptake since users could, at the very least, book a ride. Then as a bonus, they could share the ride and save money.

The other mention just came across my Google News feed for Poorsquare.  ”10 ways to eat for (mostly) free on the road”:

Check In with Foursquare

A new website called Poorsquare is helping travelers seek out food freebies or low-cost meals in exchange for checking in on Foursquare. According to the blog About Foursquare, “You tell [Poorsquare] what neighborhood you’re in, what kind of offers you’re looking for, and whether you’re hanging out with friends (they come in handy for acquiring friend specials). It spits out a list of the freebies in your area that are yours for the taking just for checking in.”

The service is available in 85 U.S. cities, plus London. If you don’t mind advertising your whereabouts on social media, it could prove an easy way to obtain some tasty handouts. A Poorsquare iOS app is in the works, too.

Unfortunately, our site looks like crap (entirely our fault!) and no one realizes we redid the entire service as a mobile web app (also mostly our fault).  We should probably mention this on the web: open poorsquare.us in your smartphone browser and it will be a lot more magical.

They say you can’t manage what you can’t measure.  For the NYC Big Apps competition this year, I’d like to develop a “noise map” of NYC and measure the number of honks over periods of time in different locations complete with recordings.

Earlier this year, New York City decided to remove all the “Don’t Honk” signs.  A paragraph jumped out at me:

But interviews with officials, residents and cabbies suggest that it is virtually impossible to tell whether honking has decreased in New York City, or to gauge how effective the signs have been. Many said they believed honking had been curbed somewhat, but wondered if they had simply stopped noticing it, allowing the horn to become the ambient soundtrack of their days.

I call BS on that. It is not impossible, it just hasn’t been tried yet.  I hate honking with a passion (more on that in another post) and right now what’s needed is a general way to measure traffic noise.

According to NYC Big Apps, one of the options is to produce a new data set that the city can use:

Generated Data must provide a significant benefit to New York City residents, businesses, visitors, and/or government. Generated Data may include, but is not limited to the following: 1) original data that is collected directly through the Application’s user input or behavior, 2) original data that is collected by the Application’s creator(s) from direct observation, including through the use of scientific equipment and sensors, and/or 3) original data that is collected by the Application’s creator(s) from direct calculation or computation. Application creator(s) may not pay any individual or organization to gather or submit data. Data-gathering and data quality-control techniques must be fully disclosed on the Submission’s Project Page. The Sponsors and Administrator reserve the right to assess and verify the accuracy, reliability, and/or validity of any Generated Data.

Here is the concept, broken into 4 parts:

1. iOS app that captures noise:

This would be a “data collecting” app (so, not much design) that one would put on a windowsill, for example. The app would passively record traffic noise.  It would be calibrated with an acceptable decibel level as normal.  Any spike (ie, a horn or a siren, etc) would get captured to the cloud along with the location, timestamp, and other meta data.  It would also count the sounds in a backend database.  It’s basically a sound activate recorder, but it would require some calibration and would actually record the audio files in a more useful way.

The app should buffer 10 seconds all the time, and monitor the waveform.  

A horn honking looks like this. It is a distinct rectangular block that goes way above the regular din of traffic.:

Just using decibel levels would probably be more than adequate.  Then it would bracket that noise plus/minus 3 seconds where the middle is the spike.

The ultimate intention of the app is to get it into many iPhones around the city to capture many locations and times so we can start to get a realistic picture of NYC noise.

A website and back-end to classify captured noise:

The above app will be importing many isolated sound files of noises (honking, ambulance sirens, etc) and the classification will be done automatically or through crowdsourcing/outsourcing.

Automatic classification:

Assuming we have a sufficient number of isolated, classified sounds, the system could compare incoming audio files to known sounds and classify with a high degree of confidence. With audio comparisons you could check one 1-second WAV file against another that is “known” to be a taxi honk and determine if there is a high likelihood they are the same.

Manual classification:

The remaining sounds would be categorized by a human ear fairly quickly through a simple interface on the web - think a play button followed by several options like “bus horn”, “siren”, “car honk”, “truck accelerating/decelerating” “other”.

In fact, I already did the above in an even more manual way.  I recorded 19 hours of continuous audio from my windowsill to MP3 and sent it to an outsourced worker to isolate each noise and classify it according to a list I provided.  The MP3 file had a start time of May 25, 2012, 4:50pm - so all the audio time-stamps would be offset from that.

This is what the WAV file looked like:

My virtual assistant extracted 1075 individual noises. He saved MP3s of the files in folders, named with timestamps.  Many sounds were just 1-3 seconds, but sirens and other longer sounds could go for 20 or 30 seconds.  With these convenient recordings, you can quickly listen to Frederick Douglass Circle (at 110th and Central Park West) and hear the mayhem that occurs daily.

He also logged the data into a Google doc.

The data is only partially accurate because the classification wasn’t great, but the idea is there.  We can start to create histograms of the various types of honks in various locations like the one below.  

Ignore the “16” because there was no data for that hour.  The rest makes sense: 5pm and 6pm are particularly loud times. Note that it never drops below about 40 loud noises per hour.  If you imagine a student trying to study for a test or someone trying to get some sleep, you can quickly see why a horn blaring nearly every minute can sap the city of quality of life.  It is something we should be measuring and those measurements should be directing DoT and TLC policies and NYPD traffic enforcement.

Public API of NYC Noise:

All of the above will feed into a publicly accessible API that could be queried by:

• GPS/location + radius
• Type of noise
• Date/Time
• Aggregate by cohort (hour, day of week, type of noise, location, etc)

The API would also deliver the actual recordings so users could quickly create sound maps as well to hear what an intersection sounds like, for example.

This API will ultimately be what is submitted to NYC Big Apps as it would represent the first data-set that measures NYC noise.

Imagine the applications of a Noise API:

I’d create a simple “noise map” website using the noise API that would allow users to explore NYC through noise.  This would merely represent a use-case of the API.  There are many, many other uses of the data that I can imagine, and I know the ingenuity and innovation of NYC techies will find even more interesting ways to utilize the data.

• Add a “noise” layer into Zillow and other rental maps
• Compare locations with “don’t honk” signs to places without them*
• Identify ambulance routes by tracing the “ambulance siren” noise type
• Identify poorly designed intersections that have higher than average honking
• Compare high honking locations with taxi GPS data to (possibly) identify honking taxis
• Use the “bus honk” type to identify buses that are honking inappropriately
• Play sounds of honking loudly at a community board meeting

* NYC has taken down the “Don’t Honk” signs saying they don’t have an impact…. though no one actually knows if anything has an impact because no one has measured traffic noise like this.  (311 complaints are statistically insignificant and represent perhaps < .0003% of the actual honking that occurs.)

What do you think?  Are these the rantings of a lunatic who is too sensitive to noise? Maybe noise is a “fact of life” in NYC.  Or is honking an issue that plagues the millions of residents who don’t own cars but are disproportionately impacted by the magnitude of noise generated by a handful of impatient drivers?

Update: I posted this idea to NYC Big Apps

We went home with the second place prize: $10k in cash!

A little backstory

Campbell’s put out a hackathon brief that basically said “build something that makes it easier for people to put dinner on the table” with some massive cash prizes.

I teamed up with some of the usual suspects - Alex, Aaron, Toby, Mike, and Rory - and we worked on a web app called Cheftacular.  (I will include our demo video at the bottom once I get the OK from Campbell’s.)  Obviously I cannot overstate how awesome my team was/is. Also, this was a great reason for us to work together and hang out over a few weekends.

The Odds

I learned that there were over 150 submissions for the initial entry round, which was just a text-only explanation of your hack and how it answered the “what’s for dinner” question.

30 semi-finalists were selected to actually build their hack and submit a 2 minute video demo of the hack in action. That was where we spent the bulk of our time. I think we had about 2 weeks.

Then, of those 30, 10 were selected as finalists to go to Google HQ and present, in person, a 3 min pitch of the hack to 3 judges - Chris Pirillo, Rachel Sanders and Campbell’s CMO Mike Senackerib.

So on Friday, four (out of six) of us brought the heat to Chelsea Market, my old stomping ground when I was a stand-in for Bobby Flay on Iron Chef America. (The Food Network and the studio is in Chelsea Market.)

Only 6 teams showed up for the event, giving us even better odds. But the teams that came were all from out of town.  Hawaii, Portland, Milwaukee, New Haven… kind of crazy.  We were the only ones from NYC. Rory and my 20-block trip was obviously the least interesting of the group.

Our Presentation

I had been prepping for our 3 min pitch for a while.  This isn’t my first rodeo but I was pretty nervous beforehand, partly because we hadn’t actually seen any of the teams or hacks or the people running the competition.

We had given them a powerpoint deck beforehand but decided it would be a better pitch if, in the words of Bill O’Reilly, we did it live.  There were a few brief meetings and we got the OK that we could just go to the website and do a live demo after the opening few slides.

I’ve been to a lot of presentations.  Usually time is kept by the MC on an iphone. Campbell’s had an official timekeeper, which was fine, except she was using the loudest, beepy-est 1980s timer ever! It looked something like this and beeped while being set.

I am distracted by loud sounds, so maybe I’m the only one who noticed.

Aaron and I learned two very important things during the presentation:

1) Check the 2nd screen.  When you hook up a laptop to a projector, sometimes it “mirrors” and sometimes it uses the projector as a second monitor.  So you might see something on your laptop but it doesn’t show up on the projector… so the judges don’t see your demo!  Yeah, that happened and took us a few seconds to realize.

2) Zoom out!  CTRL+- or Command+-.  Once we got Chrome on the projector, it was way too big. Instead of fiddling with screen resolutions, just zoom out and you’re golden.  Again, took us a bit to realize this, and with Ms. Beepy Timer 10 feet away, I didn’t want to take a chance that we’d run out of time.

I stumbled a bit, got a few laughs and got to show off more in the 3 min Q&A afterwards.

Other Teams

The other teams were really impressive.  This was why I was so nervous. At a hackathon, at least you can see who you’re up against, hear their ideas, etc.  With this, we had no idea.

After nearly every presentation, Aaron and I looked at each other, eyes wide, and shook our heads like “yeah they’re gonna win”.  They were really solid ideas.

Whoa, we won

We were ecstatic to win second place!

FWIW: The team that won were two guys who run a creative agency, so we didn’t have a chance against them.

During the judge deliberation and after the announcement, we got to hang out and talk to a lot of friendly Campbell’s folks.  Finally, we got to put some faces to the people who ran the competition.

Also, I made it into Campbell’s official twitter feed:

Unfortunately, it was a bit of a blurry picture that looks like I’m in a bathroom, but whatever. I’ll take it.

This is where I was standing:

The goodie bags were, indeed, full of goodies, including an apron. (This pic is from a tweet by Alan, who built “Recipe Remixed“… which I thought was the coolest hack. Like Github for recipes. He came from Hawaii!)  

Not gonna lie, I enjoyed the hell out of that cream of mushroom soup that night.

We had a fun time doing this hack.  Looking forward to the next one…

You may have heard about this story:

The construction worker rescued from thick, chest-high muck at the Second Avenue Subway construction pit after being stuck for four hours today wanted just two things — a comfy bed and a cold beer.

— via NY Post

But there’s more. After being stuck in mud for 4 hours, the guy goes to the ER at Cornell-Weill where he waits - literally, sits in the waiting room, right after being rescued - for what appears to be 15+ hours.

He plans to kick back with a cold one as soon as he is discharged from the hospital.

“I’m going to have a beer and relax when I get home,” he said.

Until then, he’d settle for a room in the hospital, which was so overcrowded he was stuck in the ER all day and into the evening yesterday.

“I haven’t slept since yesterday morning. Probably the adrenaline,” said Barone, who was rescued at 12:40 am on Monday.

“I’m keeping myself busy [in the emergency room]. Being here is like watching Scrubs.”

His wife Candy said he always keeps a positive attitude.

“He’s always in a good mood,” she said.

But the accommodations aren’t helping, she said.

“What do I have to do to get a room? He hasn’t rested at all!” she said.

— via NY Post

I think they buried the lead here. I think it should have been more like:

Guy waits 15 hours in ER after spending 4 hours in muck

A rescue effort 100 feet underground involving 50 fireman and rescue workers takes a quarter of the time it takes for the guy who got rescued to see a doctor at the ER.

If you get stuck in muck at the 2nd Ave Subway construction site, you’re in luck.  But we can’t help you if you wind up stuck in the ER waiting room.

Clearly a NYP headline writer could do a better job here.  But you get the idea.

I love dropbox but when you are sharing folders with a lot of people, things can go wrong.

A few months ago, one member on a 22-person shared folder in my office decided to clean house and delete some files from her computer.  She didn’t realize the folder was shared and I quickly learned the hard way that she had deleted a bunch of files from my mobile folder.

No big deal. That’s why we use dropbox, right?

Oh god, wrong.

As it turns out, there is no “revert back to this date” function.  This is the feature I am begging you to build.

Instead, you have a “show deleted files” option.  Problem is, the files that were deleted were plentiful - like, in the thousands.  And “show deleted files” literally shows you EVERY file that was EVER in the folder and was deleted.  So a TON of files that I intentionally deleted were showing up here. This is virtually impossible to navigate if you have many files.

Next, you can’t undo a particular person’s changes.  So I went hunting for some option or activity log that would have said “Jane deleted 1,564 files - timestamp 3:45pm”.

In fact, there is “events”, which does this in a shoddy way by a) only showing MY activity (not helpful) and b) not providing any kind of action to undo or batch undo those changes.

So I had to go through the deleted files list and one at a time go to the file’s “version history” and restore to the previous.  Ugh.

Fast forward to just a few weeks ago. A guy in IT wanted to wipe clean a laptop that happened to be one of the shared people on that same dropbox folder.  Guess what he didn’t realize as he MOVED all the files in the entire dropbox folder to another folder?  That’s right!

Once again I found out the hard way - read: the entire work folder missing every single file - that something had just gotten funky.  I was able to see whose account was the last to access the files and then went over to IT and asked politely WTF?

The correct way to deal with this would be if Dropbox had a “revert to x date” feature that simply restored everything, not just individual files.  Better yet, it should also have a “undo everything THIS USER did back to X date” - that would have been one step to fix this mess.

Instead, we had to literally recopy all the files (they weren’t deleted, thankfully, just moved to a different, shared network folder) back into the dropbox folder.  Then I told a few people “hey, everything you did in the last 5 hours is gone, yer welcome!”

Maybe Box is better at this. Or maybe Dropbox for business is meant for this stuff.  Nevertheless, this seems like a really obvious problem that other people can easily have that should have a fix that is as easy to use as dropbox is to totally mess up.

I spent last weekend participating in the Financial Empowerment Hackathon, put on by the Office of Financial Empowerment (OFE) of NYC.  Originally, I was going because my coworker had a neat idea for a microfinance/savings app… but he didn’t make it!  And neither did another coworker who was going to do design.

So I headed down to 155 Water St in Brooklyn to learn all about how the OFE is helping thousands of NYers get out of debt and improve their financial situations.

These types of hackathons are pretty interesting because the participants are really meant to learn about their problems (ie what the OFE counselors have to deal with) rather than work on, say, an app that does something with music or ecommerce.  In a lot of ways, these are harder hackathons because they require a lot of learning and you have to leave your assumptions at the door.

I learned that the average income of OFE clients - people who come seeking assistance on how to get their finances on track - is just $15k.  Many are single moms.  Many do not have regular access to computers or internet or don’t check their email that regularly.  They may not have a smartphone.  They are often super busy juggling jobs and kids. The counselors - those who meet and help clients understand their bills, their options and debt and other things - have a lot of trouble with reminding clients about appointments or with clients showing up without necessary papers.

I thought of a few ideas while listening to the counselors explain their pain points and answer questions. I pitched them to a few people to gauge their reactions and several seemed to form a consensus around one idea in particular: a text that gets sent on payday to remind you to set aside some money.

Pay Yourself First

Here’s what I scrawled down:

The view:

The organizers were wonderful and fed us amazing food. I commented many times to them that for a hackathon to be awesome, just get a lot of food and have internet that works - you’d be amazed how often those two requirements aren’t met.

Here’s my full pitch:

It’s hard for OFE clients to save their money.  It sounds like a common theme is that clients have a hard time setting aside money when they get it.  Millie, one of the counselors I spoke with yesterday, said she always advises PYF – pay yourself first. And that’s what this hack is all about.

So when a client gets their paycheck, that’s the best time to remind them of their financial goal and that they should set aside some money for it.  Pay yourself first sends a text message on payday reminding the person to set aside money for the goal they set up. They can then respond how much they were able to contribute that period and we keep track of that. 

This is the form for setting up a client.  It’s meant to be filled out while the counselor is having a session with them.  So lets say I’m the counselor and you, Will Tucker, are the client.

Let’s fill out the obvious stuff. I’ll say your paycheck amount is in the 500-750 range and you’re expecting your next paycheck this Friday, and you get paid bi-weekly.

Now, what’s the main savings goal? Lets say you want to save for an emergency fund and need 600 to do it.

That’s it.  You’re now set up and when payday comes you’ll get a text message reminder. 

We’re not going to wait until Friday so I’ll just send it now so you can see how it works.

»  10% of paycheck // for the emergency fund.

It’s also fully interactive. So Will got this text message but then he can text back how much he was able to save.  In this case, $50.  

» how much you saved, contributions, goal

If you ignore it, it’ll just be saved as a payday that had no contributions. This does two things: tracks contributions for the client, but also the help counselors identify clients who are falling behind.

Now, on the admin page, the counselor can review how their clients are doing.  Name, the goal and amt, and the # of paydays that happened since they started, and the # of contributions they made.

This is a first attempt but I thought that the % of the time a client contributes was the most important factor – to build a routine where saving, even small amounts, is a regular thing.

For counselors, there’s a “send appt reminder” button that’ll basically identify all the clients who have contributed < 50%  of the time – these might be clients who need an extra push and need more face time.  Click this and it’ll send a text to all of them with this (it’s editable).  If they respond “yes”, we’ll send their info to OFE and, hopefully, a counselor will call that client in a few minutes to pro-actively schedule an appt.

This is a PHP application built using jquerymobile and Twillio for texting.  Most clients have access to cell phones with basic text messaging – so we’re using that as the primary means of communication. But they’re also very busy, so on the go, they can get a quick reminder away from the computer.

There’s a lot of opportunity for next steps.  I’d love to see users get on this and figure out what can be tweaked to get better outcomes.  This is fully functional, so you can sign up right now and get a text.

OFE is doing great work and I think Pay yourself first would further the mission by empowering clients.

There were only about 15-20 people hacking and 7 teams presented. But the top prize was $1,000 with two $500 runner-up prizes, so the odds were definitely good.

I liked that this was almost half-hackathon, half-conference. Everyone emerged with a much deeper understanding of this NYC service that I don’t think many of us knew much about beforehand.

Here are some instagram pics of my presentation (thanks Yangbo):

Pay Yourself First was a runner up! (That’s me in the middle with the rest of the Village of the Damned.)

And here’s the NYC.gov press release about the event.

Meanwhile, it sounds like the OFE counselors may actually use some of these hacks!  Normally most hackathon projects are forgotten about by Monday morning, but this would be a welcome change.

# of Hackathons won or placed in: 7

Still… sad to be missing out on StartupBus and Launch’s Hackathon :(

Whenever I hear about a new idea someone has for a startup or a product, the first thing I usually ask is “what other products solve this problem?”  I don’t call them competitors because these are just ideas or beta products and because for most ideas, it’s safe to assume that the market is big enough to accommodate another player.  If not, it’s probably way too small a market.

So the recommendation I make is for the founder to assume that 1,000 equally smart and motivated people have come up with the same idea or at least have experienced the same pain.  And that the solution already exists.

I literally “will” it into existence.  I don’t know there is a product that does X or Y but I have to assume there are at least 10 products that come really close to nailing the solution to the problem.

The image I usually conjure up is this clip from one of my favorite movies growing up, Bill and Ted’s Excellent Adventure.  They’re walking at the police station and decide that after their adventure, they’ll need to come back to that place, steal the keys and put them behind a sign… in order to break in, at that very moment.  They come up with the idea and then presto, the keys are literally sitting behind the sign.  This isn’t exactly what I’m talking about with startups but I think it’s sort of analogous.

So the first thing you should do is assume there are already products that do what you want done.  Put the keys behind the sign.  Then find those products by assuming they must exist. If you don’t find them in a few google searches, you’re not searching hard enough. After all, they exist.  That is the assumption.

Next, go and use them. All of them.  If one of them works and solves your problem adequately (say, 80% of the way but has room for improvement) then my recommendation is usually to celebrate the fact that someone else spent all that time and built an awesome product. Your problem is now solved and you don’t have to waste your time and money trying to build a business that is likely to fail.  Instead, you can spend your time looking into other opportunities.

If, however, all these other products just don’t nail the solution very well and you’ve corroborated that with a few dozen non-friends who have the same problem and feel the same way (ie, early customers or beta users), then maybe there’s a startup in there after all.  

Usually there are services and products that nail 60 or 70% of a problem but fail in a few specific ways.  In that gap is where you can focus most of your efforts.  That’s called differentiation, and it’s where you can often find the ways in which your thing is better than other products for a certain problem.  It can often define how you pitch your startup or how you develop it or what market you go after.  The earlier you identify what your differentiating features are, the better.  It’s how you’ll pitch people about your business.

But it all starts with an assumption.  Assume we live in a multiverse. San Francisco, Chicago, Austin, Boston, etc all have doppelgangers of the most brilliant tech minds in NYC, including you. They are all experiencing similar problems and ideating at meetups late into the night, going to hackathons and emerging from accelerators.  Assume that each city has their own cool new startup for whatever it is you’re doing.  I swear, every city has their own “catering on demand” or “social photo sharing” service.  Chances are it’s a crowded market just in your city.

Your job is to find it and know everything about all of these products and services and the people behind them.  If they failed, or aren’t around anymore, why is that?  Reach out to the founders and learn what you can.  

You’ll probably find a cool product that makes your life easier. Easier because it solves your own problem - you know, the problem that kept you up at night. Now you can sleep! Also easier because it saves you from building something that never needed to be built in the first place.  All of that money and time you were about to spend can be directed to an even better idea that has a clearer value prop.