Of the DOGE list’s initial claim of $16 billion in savings, half came from an Immigration and Customs Enforcement (ICE) listing that was entered into the Federal Procurement Data System (FPDS) in 2022 with a whopping $8 billion maximum possible value.
According to a DOGE post on X, that number was a typo that was corrected in the contract database to $8 million on Jan. 22 of this year before being terminated a week later, and DOGE “has always used the correct $8M in its calculations.”
But for much of this week, DOGE listed the outdated $8 billion for its savings claims while linking to the termination notice with the smaller ceiling amount.
Some time Tuesday evening, the DOGE link was changed to point at the original $8 billion entry, and on Wednesday morning, the site was revised once again to show $8 million in savings — but still linked to the larger, outdated claim. The site also continues to list $55 billion in total estimated savings — the $8.5 billion in alleged contract savings and another $46.5 billion with no specifically documented source.
Spokespeople for the White House and DOGE did not respond to multiple requests for clarification about the DOGE data and savings claims shared online.
…
Just over half of the contracts touted by DOGE, accounting for $6.5 billion in alleged savings, haven’t actually been terminated or closed out as of Wednesday, according to an NPR analysis of a federal government procurement database, even though the site’s “wall of receipts” listed these items.
…
In all, estimated savings from the initial DOGE list of just over 500 contracts that NPR found to be cancelled runs closer to $2 billion, with roughly half coming from the gutting of the Department of Education, the Consumer Financial Protection Bureau and the U.S. Agency for International Development.
Archived at https://ghostarchive.org/archive/w8hLJ
The thing is that it’s difficult to even get a complete picture because they release so little data.
One screenshot I saw had the PIID in the shot so I looked it up. Doofus indicated he saved $8.2M but that was the total amount for the outlay.
The contract began in 2022 and had already paid out 80% the total $8.2M, meaning only $1.64M was actually “saved”.
So I know for a fact that the values they’re reporting aren’t correct because they’re looking at these contracts and having no clue on how to actually read them.
As for the SSN thing where we’re paying 150 year old social security. There’s likely another table with “in question” that has to be joined to get an actual number.
When I worked for my state this is how we had some data. A master table that you then had to join like five or six exception tables to remove the “questionable” entries from the master.
And on systems like AS400,
EXCEPTION JOINis a thing for a reason because the OS SQL engine has special ways to optimize these kinds of joins. So it’s likely this is the case here.TIL
EXCEPTION JOIN. I thought the SQL dialect I usually use at work for data warehouse queries (Presto) didn’t have anything like this, but it does, and calls itEXCEPT: https://prestodb.io/docs/current/sql/select.html#union-intersect-except-clauseGood to know. Beats my usual approach of using
WHERE x NOT IN (SELECT ...). I’m mostly a front end developer so these things are outside my comfort zone sometimes.When I worked for my state this is how we had some data. A master table that you then had to join like five or six exception tables to remove the “questionable” entries from the master.
At my workplace, we’d have a data pipeline (think something like Apache Airflow) that pulls the master table once the daily partition lands, joins the exception tables, then produces a “clean” output table which is the one that people would actually query. At a previous employer, we would have used a materialized/indexed view (we used SQL Server for both OLTP and OLAP). Is that not common in government?
Of course it doesn’t. Lie and project. That’s all they know. And their base will slurp it up. Especially without an education.
