By complete accident I just discovered that pressing Ctrl-K in the compose email field in Gmail will bring up the Insert Hyperlink dialog. It’s not in the Keyboard shortcuts dialog (visible by typing ‘?’ outside a text field), I don’t know why.
This will save me at least an 20 minutes a week of taking my hands off the keyboard to reach the mouse.
I love using Rapportive, the browser add-on that brings contextual social information to your Gmail inbox. It’s really helpful to see the face of someone that you’re contacting. It’s also extremely convenient to be able to see their latest tweets, Facebook updates and LinkedIn profile while writing an email to them.
In Kontagent I use Salesforce extensively to record information about relationships with clients, and find myself constantly switching tabs between Salesforce and Gmail to write my emails. I also use Salesforce in tandem with Zendesk and Google Docs, and would love for those products to be integrated more closely. I decided to kill two birds with one stone and designed a Rapportive plugin (or Raplet) that pulls information from Salesforce while I’m writing emails and also uses that Salesforce info to pull other information from Zendesk and Google Docs.
You must host the code on your own (HTTPS strongly recomended) server in order to use the Raplet. This is because you must hardcode in the security credentials to access your Salesforce instance before using it.
The code is available to download here: https://github.com/conornash/CRM-Raplet
I’ve found the Raplet has already saved me the time it took to make, especially with larger organizations where I may have a number of conversations going at one time with different people. I’ve also made variants of the Raplet for different departments in Kontagent showing information that is more relevant to them, which has been received well too.
Please let me know if you have any questions about using it by sending me an email at raplet-help@conornash.com
I spotted a nice tool from Google last week that I hadn’t heard of before. It’s called Google Correlate and it allows you to see two fun little pieces of information – one of which is a way to ‘prove’ (i.e. trick) people into thinking that there are patterns present when there are none.
Firstly, you can see how correlated search terms are to each other by frequency over time. This is not very interesting as a particular search term is most likely to be correlated with variants of itself. For example, if you input ‘dvd’ as a search term the top correlations are ‘dvd burner’ and ‘dvd recorder’. Maybe this would be good for picking under-utilized AdWords keywords.
The second aspect is more interesting. You can see how correlated Google search terms are with real-world data, either time series or information about US states. The time series is cool, but being able to correlate any State-oriented data with Google means you are almost guaranteed to generate something that a tabloid paper can run with. Here are some interesting correlations I’ve found while playing with the tool:
Annual Rainfall of US state (in.) <-> Google search disney vacation package: 0.9093 (link).
Literacy Rates in each US state (% of population) <-> Google search olympics: 0.8897 (link).
Southern US states (as defined by US Census Bureau) <-> Google search crape myrtle: 0.8982 (link).
Populations of US States (inverted) <-> Google search gatorade player of the year: 0.8963 (link)
US States that I have visited <-> Google search virgin america airline: 0.8617 (link)
Do illiterate people hate the olympics? Does rain make you want to go to Disneyland? Do I fly with Virgin America everywhere in the US? The answer to all of these is, of course, no. Even so these kinds of correlations are often presented in media as ‘evidence’ of a subjective opinon that has no basis in fact. I feel the best way to fight this attitude is with education and access to the data cited. Until then I’m sure more than a few newspaper headlines will be generated by correlations such as the ones above.
I’ve recently accepted a new position at Kontagent where I will be a Data Scientist / Account Manager so expect even more data-driven blogposts in the future. I made the decision to move to Kontagent when I saw the super-smart team they had assembled and the data that I would get to probe and interpret. Heyzap has been a huge part of my life for the past 18 months and I wish them every success in the future – and I’m sure I’ll see them around SF regularly.
At Kontagent I’m working hard with the Mobile team to make a truly enterprise-class mobile analytics platform that builds on the 3 years of hard work the Kontagent engineers have already put in to the Web platform. With over 125 million MAU tracked on our system we’re no stranger to working at scale. I’m also excited to announce that we are accepting a limited number of partners into the Alpha Tester Programme for Kontagent Mobile!
If you are interested in getting on board the Kontagent Mobile Alpha, please let me know in the comments and I’ll see if I can wrangle you an invite.
Great companies are made and lost in Silicon Valley based on the quality of their engineers. One great engineer is often a better asset to your team than twenty mediocre engineers, and one bad engineer can actually make overall productivity decline. Being able to attract and retain this top 1% of the talent pool of engineers requires companies to offer six figure salaries and extremely competitive benefits such as the (so-called) “unlimited vacation time”. It also requires less-quantifiable benefits, such as being a company that other developers respect and admire.
The quantifiable factors are easy to measure, naturally. But to measure the respect and admiration a company garners from developers, especially relative to each other, is something that has no authoratitive answer. Edwin Chen has made the first attempt at a rigorous, data-driven answer by measuring the number of mentions of the six major tech companies in the titles of submissions to Hacker News. The graph of “relative interest” over time is inserted below:
Graph above copyright Edwin Chen
Engineers are happy to admit being drawn to quantifiable benefits. However, if you mention to certain engineer that they may have taken a position because “they just heard good things about it” or that they only want to work there because the place has good PR and they might get a little sensitive. However, once that PR becomes Hacker News submissions that can be easily quantified engineers finally have the authority to say to their friends “Oh yeah, I’d love to work for Facebook. Have you seen the rate of increase of their RI (Relative Interest) on Hacker News over the past 6 months? Those guys know what they’re talking about.”
I would love to be a data scientist. It’s a brand new field, a new frontier of understanding information at volumes that society has rarely tried to grapple before. At least, that’s what i believe data science to be.
Appearing in the media about a year ago, possibly as the secret love child of Drew Conway and Hilary Mason, the term ‘data science’ has not been widely welcomed by many statistical or IT professionals. This can be at least partly attributed to the lack of a clear definition of the subject. Pete Warden rebuts this and most of the other criticisms leveled at data science in a clear defense of the term.
I would go a step further, and say that the term is a fantastic one to attract students and younger people to the numerate sciences, statistics in particular. The image of statistics is commonly seen as dour and sterile. Data science is promoted in the wider media as vibrant, easy to understand and attractive. The worlds of mathematics, computer science and statistics have been given a great opportunity to show their best sides with data science and should be actively encouraging it as a vehicle for the promotion of more “clearly defined” fields of study.
R is widely lauded for the wealth of packages that are freely available through the CRAN network. Since all use R as the environment in which to work, it could be assumed that data processed by one package could be natively read by every other. However, at least in the field of social network analysis, this is not true. It is increasingly difficult to combine use of multiple packages on one network dataset as each has it’s own preferred data format.
The igraph and sna packages are widely used and perform different roles in social network analysis. However in order to use them in tandem you must keep two separate data objects stored in memory. Keeping track of the same graph in two different formats is inconvenient at best and can easily lead to inconsistencies in analyses. This is not to mention the memory restriction that is imposed when you are dealing with large datasets.
Social network analysis is booming in popularity, in part to the increasing availability of large volumes of data in easily interpreted formats. Reproducible research is central to the SNA community being able to understand and iterate on results made public. Network data analysis has the potential to be one of the fastest growing fields of study of this decade – fragmentation of data formats for network data could stifle this growth or worse, put the results in proprietary formats unreadable by the community at large.
