Is Palantir's Crystal Ball Just Smoke and Mirrors?(nymag.com)
nymag.com
Is Palantir's Crystal Ball Just Smoke and Mirrors?
https://nymag.com/intelligencer/2020/09/inside-palantir-technologies-peter-thiel-alex-karp.html
3 comments
I'm not sure Palantir has any technology moat whatsoever. All of their tech is either open-source off-the-shelf or licensed from other vendors. A company driven 100% by sales/who-you-know! It's surprising they keep wearing the fig leaf of pursuing non-government work, considering how much their commercial clients churn.
I interned at Palantir some time ago. I had essentially no idea of what their software did until I had a demo at the onsite, and it wasn't really clear what the product was really for until I started working there. The product is simultaneously very powerful, but also incredibly underwhelming and doesn't live up to Palantir's mystique at all. This sounds contradictory, so let me explain:
Palantir builds data visualization software. They repeatedly claim that they do more than data visualization, and while that's true to an extent, easily 80% of what they do is data visualization. The key thing that Palantir does is their ability to ingest diverse sets of data sets from a variety of sources (often crufty government databases) and build relationships between this data. Say you're tracking down pill mills. Transactions on Schedule II drugs might be in one database, and criminal records related of pharmacies are on a different database. Palantir can ingest both data sets, and let people browse Schedule II drug transactions at pharmacies with criminal connections. And then do things like filter for people who bought the same drug twice or more in one day, a big indicator of nefarious activity.
This is pretty boring in my opinion, but the reality is that this workflow is often replacing things like browsing rows in Excel. I think the company's value is real, this tool does make it a lot easier to do a lot of tasks that involve making correlations between different sources of data.
The issue with demoing Palantir is that there's no one product. Something like half the company's engineers works on customer-specific customizations and UIs. Here's a couple demos that showcase a couple versions of Gotham:
https://www.youtube.com/watch?v=jZug8kV28RU
https://www.youtube.com/watch?v=0DzIQ-ygQV4
https://www.youtube.com/watch?v=XCR_LBzur-k
https://www.youtube.com/watch?v=_Q8bwhAW2Mg
A more in-depth explanation: https://www.youtube.com/watch?v=XyDEHCcQ_bk
Overall it was an interesting company and I liked my time there. I said "the product is boring and does not live up to the mystique at all" but that's referring to the user-facing product. The underlying technology is actually pretty cool. Fitting all these different sources of data into one data store and building effective abstractions for developers to retrieve that data and build useful features on top of it is a challenging problem.
Palantir builds data visualization software. They repeatedly claim that they do more than data visualization, and while that's true to an extent, easily 80% of what they do is data visualization. The key thing that Palantir does is their ability to ingest diverse sets of data sets from a variety of sources (often crufty government databases) and build relationships between this data. Say you're tracking down pill mills. Transactions on Schedule II drugs might be in one database, and criminal records related of pharmacies are on a different database. Palantir can ingest both data sets, and let people browse Schedule II drug transactions at pharmacies with criminal connections. And then do things like filter for people who bought the same drug twice or more in one day, a big indicator of nefarious activity.
This is pretty boring in my opinion, but the reality is that this workflow is often replacing things like browsing rows in Excel. I think the company's value is real, this tool does make it a lot easier to do a lot of tasks that involve making correlations between different sources of data.
The issue with demoing Palantir is that there's no one product. Something like half the company's engineers works on customer-specific customizations and UIs. Here's a couple demos that showcase a couple versions of Gotham:
https://www.youtube.com/watch?v=jZug8kV28RU
https://www.youtube.com/watch?v=0DzIQ-ygQV4
https://www.youtube.com/watch?v=XCR_LBzur-k
https://www.youtube.com/watch?v=_Q8bwhAW2Mg
A more in-depth explanation: https://www.youtube.com/watch?v=XyDEHCcQ_bk
Overall it was an interesting company and I liked my time there. I said "the product is boring and does not live up to the mystique at all" but that's referring to the user-facing product. The underlying technology is actually pretty cool. Fitting all these different sources of data into one data store and building effective abstractions for developers to retrieve that data and build useful features on top of it is a challenging problem.
Thanks for sharing some thrilling insight! From what I have gathered my own conclusion was pretty inline with what you say. That the real power is behind the scenes at the database level and customizations of datasets. When I first saw the UI (pretty recently) I was underwhelmed. I'm curious if you can elaborate some more if you think that the UI can be changed/improved without changing the fundamentals or is it tightly coupled?
It seems to me that they have one product per customer. I'm curious how scalable it is I guess.
Aside. I became very interested in Palantir. Much because I liked Alex Karps philosophy in that this technology is ineveteble, controversal or not, and that it's up to regulators to scope the use of it. If not Palantir some other company down the line will do the same (if not already) and personally I'd go with team Karp.
It seems to me that they have one product per customer. I'm curious how scalable it is I guess.
Aside. I became very interested in Palantir. Much because I liked Alex Karps philosophy in that this technology is ineveteble, controversal or not, and that it's up to regulators to scope the use of it. If not Palantir some other company down the line will do the same (if not already) and personally I'd go with team Karp.
https://www.businessinsider.com/activists-protest-palantir-a...
I'd love to believe Palantir organised the protests, but sadly I don't think so.
We got idiots over dystopian Blade Runner :(