I think the article misses an important point: SAS is a generational thing, and the influence of generation that cut their teeth on SAS and VAX mainframes in the 1980s is waning. There's a reason why the SAS ads you see in the airports are all targeted at management--because that's where all those people work now. And soon they will retire.
They would never dare advertise to the poor analysts who are forced to actually use SAS to get work done, because we know what turd it is. I do not think I exaggerate when I say that not a single up-and-coming statistician in this day and age chooses expensive, obfuscated SAS as the language with which to hone their skills. It is truly the worst language I have ever had the misfortune of being forced to learn (by an old professor, who knew SAS and refused to learn anything else.) Somebody back me up here.
There are a few areas where SAS is unrivaled--it can crunch huge data and exploit parallel processing better than other off-the-shelf solutions. That feature gap will close over time as R and other open tools rise in popularity. And there will always be a niche market for SAS programmers because many companies have huge libraries of SAS code which run their credit card/oil rig/whatever analytics, and starting from scratch is not an option. But SAS's dominance is unquestionably on the decline, and a lot of it has to do with the fact that next generation a) isn't being trained on SAS and b) doesn't want to be.
Yeah. It used to be that SAS, like Oracle, was the only place to turn to when you had very large quantity of data. The internet has added multiple orders of magnitude to large, and people now use MR frameworks.
I haven't seen SAS code before. I'll take your word that it's ugly. However, if SAS can adopt new standards, I hardly see them waning. Companies will always be thirsty for better ways to understand their customers, and now with the ease at which new internet companies are started and more people become connected, the demand for those services is only going to increase. If SAS can continue to outperform by investing in R&D and adopt new ways to bring in fresh talent (ie use more open-source tools) to keep them ahead of the curve, they'll do just fine.
> The company shifted course earlier this year and modified its software so programs written with R work seamlessly with SAS technology.
I agree that getting them while they're young - like McDonalds - is an effective strategy, but if you can swap out the language for another front-end, does it matter? (it's a big "if").
The origin is a stats system. For a long time, they were the only stats package in town that was good at working on data that was larger than available memory. This is still mostly true, but people's definition of large data has outrun what even SAS can offer and people are falling back on MR frameworks.
They make their money with BI. It's a language, database system, ETL, and set of premade programs that are all about BI and analytics. They have stuff built out for large clients, so if you are eg a university or a large manufacturing concern or a bank or an ibank or a casino, etc, there are large bodies of preexisting functionality.
I really enjoyed that article. The active M&A in business intelligence software companies shows just how big Big Data is for business, and how big the market is for new startups to tailor specific techniques for niche industries. However, now that the secret is out, its going to be more difficult to prove that your secret sauce is superior than your competitors. The winners in this market will be those who can deliver results fast, and be nimble enough to account for the elastic nature of Big datasets.
BI software and enterprise software in general is a huge opportunity for startups - I know PG and YC had it as one of the RFS listings at one point.
I'm not sure that the winners will be those can deliver results fast and exhibit quick adaptability. In general, I agree that the Hacker News community is more than technically capable enough to do better than the big BI offerings. But I've been surprised (or shown my naivete if you prefer) at how much "old school" networking still accounts for sales in this space. To succeed, you are aiming at mid-level and higher execs who are used to (seriously) being wined and dined in a way that I think most hardcore tech companies would be surprised at.
Figuring out how to circumvent the normal sales channels is probably at least as important as whatever tech you put in front of people for enterprise software. I've seen how successful some companies are at simply pushing software out to their existing customer channel and making $$ off of that.
All that said, and I still have my eye on a little slice of that pie chart . . .
You'll probably need quite a bit of money backing your startup if you're going to be handling enterprise data to let them know you'll be around years from now. Enterprises need that assurance before they'll even deal with you.
The amazing thing about SAS is that it makes R look like a beautifully designed language. (R is great, but has a LOT of idiosyncracies related to silly type conversion defaults.)
AFAIK, SPSS mostly runs regressions.. linear, logit, clusters.. and same for SAS. Calling them "Business Intelligence" appears to me as some BS name to sound more impressive to mgmt in big firms.
One thing that's not mention prominently in the article is that while SAS provides many perks to its employees, it actually has a history of paying salaries significantly below industry standards.
Not really. IBM has a huge presence, still over 8500+ employees I think and has traditionally paid at the higher end of software developer salaries on average. And until Nortel imploded this decade, they were also in the mix as a top-paying, always-hiring type of place here.
Basically, SAS has always paid significantly less than others in the park (RTP). I think it's always been pushed as more of a lifestyle place - SAS is still privately held, so it was supposed to be a little more "sane" than big, quarterly focused IT companies.
I do know folks who have worked there and have enjoyed significant work flexibility (eg, working from home from other states) but also have worked with ex-SAS'ers who left because of what I understand to be a laid-back to the point of not getting work done atmosphere.
While cost of living is good in NC, here in the Raleigh/Durham/Cary/Chapel Hill area, it's definitely significantly higher than other places in NC. It's much cheaper than the biggest metro areas (SV, Boston, NY, DC) but after that it tends to run higher than you might think, particularly if you are shopping for homes in areas with "good" school districts.
