Andrew DeCotiis-Mauro

So many things to say but not enough time to write

@DataRepublican @JustinPDuncan @JuddLegum @elonmusk @Oilfield_Rando Why don't you just provide the queries for analysis and critique?

Yeah, you’re not wrong. The actual number is somewhere between total_obligated and total_outlays. To estimate the actual money spent, you’d have to drill down further to percent of contracts cancelled. Likewise, to actually correct for the accounting errors in total_outlays, you’d have to dive into the transactions table. This is why metrics is HARD. Regardless, since the lower bounds (total_outlays) is within 97.4% of the cumulative potential award value, I’m satisfied to call DOGE the correct metric here.

1. So @DataRepublican, who has been promoted repeatedly by @elonmusk, published this post claiming that methodology of the Musk Watch DOGE Tracker is flawed. She claims in the post below that she could not find a single contract that ended in 2024 where the outlay was less than the "Potential Contract Value." Not one. She does not have any idea what she is doing. In this thread I will provide 75 links to contracts that ended in 2024 where the outlay is less than the "Potential Contract Value," totalling $57 billion. This is a small subset of such contracts, from only three agencies. Moreover, the broader point is that canceling an option to extend or expand a contract does not save any money. The government holds these options and could simply not exercise the options. As this thread demonstrates, it frequently does not. @DataRepublican, I look forward to your corrections and apology.

@coldwintersand @JuddLegum @DataRepublican @elonmusk I'm still waiting for the query

@DataRepublican @JuddLegum @Oilfield_Rando Where are you sharing your queries.

Hello Mr. Legum, You’ve built a website that relies on a specific interpretation of “potential” in the context of federal spending it. While @Oilfield_Rando mentioned this earlier, I went a step further and confirmed these definitions myself. In the USASpending rpt.awards_search table, larger federal contracts often specify two key fields: 🔹 ordering_period_end_date (labeled as the “Potential End Date”) 🔹 base_and_all_options_value (labeled as the “Potential Award Value”) Contrary to everyday usage of “potential” (which implies something uncertain or unlikely as your website suggests), “Potential Award Value” in these databases represents a contract’s maximum cost if all negotiated options are exercised. It is not a mere “possibility”; it’s the upper bound of contractual spending authority. And in practice, contractors squeeze these agreements to their fullest extent, because why would they give up money? To prove this point, I ran a SQL query to extract the first 60,000 contracts expiring in 2024 with outlays over $1 million and with no ordering period extending beyond 2024. Out of these awards, zero contracts had an outlay less than its Potential Award Value, and the vast majority vastly exceeded their Current Award Amount. So, yes, I concur with @Oilfield_Rando - @DOGE, if anything, is underestimating their own savings.

1. The DOGE website is garbage. The claimed savings has no relationship to the data provided. And the data that is provided grossly inflates the savings from each canceled contract. So we just launched our own site — the Musk Watch DOGE Tracker — that breaks everything down

@EcZachly Cool. They need more people without SME

Going to be interviewing with DOGE soon y’all! Time to use data engineering for the good of the American people and save the debt!

@RepSherrill @threadreaderapp unroll

In the United States Armed Forces, you swear an oath to the Constitution — not a person, not a political party. It’s an oath I took for the first time at age eighteen when I entered the Naval Academy, and have taken numerous times throughout my career. I will never abandon that oath, and I will always fight to protect the men and women who risk their lives to serve our country. (9/9)

So much for a “meritocracy.” Last night at the Pentagon, President Trump and Secretary Hegseth dismantled a core principle of American democracy: the apolitical military. Let me explain what’s going on and why it is deeply dangerous for our country: 🧵

@johnastoehr @threadreaderapp unroll

34. He tried to meet the press corps’ phony moral standard, only to have it move around, beyond his reach, thus surrendering his power. In the end, his dependence on the press corps made it so he had to ask for permission to campaign. Harris isn’t asking. /end

32. Reporters like Cillizza have a bad habit of presenting themselves to voters as if they operated in their interest and we know nothing could be further from the truth. We should not only stop tolerating this bad habit. We should be hostile towards it.

@AdamRackis @msdGYT exactly why I switched like 15 years ago

@msdGYT Nah grass is green here in Mac world. Shit works and I don’t have to tinker on my computer (which is a waste of my time)

The one person I know who actually uses Linux on the desktop is a brilliant Senior Staff Engineer. Last time I saw him I said something like “ayyy still using Linux eh?” — he sighed and said “yeah but things keep breaking and I’m sick of it. I’m about to give up and get a Mac” 😂

the White House said use better programming languages bro whitehouse.gov/oncd/briefing-…

“The fundamental problem of software testing… is that software has to handle many situations that the developer has never thought of or will never anticipate. This limits the value of testing, because if you had the foresight to write a test for a particular case, then you probably had the foresight to make the code handle that case too. This makes conventional testing great for catching regressions, but really terrible at catching all the “unknown unknowns” that life, the universe, and your endlessly creative users will throw at you.” - Will Wilson

@generativist It was great when you could afford to buy an apartment complex for a $10k loan. Whenever that was

on average i assume the people who listened to the rich dad, poor dad guy are probably poorer than they would be had they only just…not done that

Join @attorneyjeremy1 and I today 11AM EST to talk about legal limit of how much wood a woodchuck can chuck . . . No wait blockchain and customer/user rewards! For a little sneak peak of an announcement to come, see the next “X” in this thread twitter.com/i/spaces/1ynJO…

This Thursday at 5 pm Eastern (2PM PST) I will be hosting my first space called “Seedstarter Money Tree”. This will be a recurring space allowing listeners to learn more about the companies offering on @JoinSeedstarter. twitter.com/i/spaces/1djGX…

@jesse_hynes Why you talking bout me

Confirmation bias has always interested me. Two people can look at the exact same thing and interpret them in polar opposite way. It’s always interesting to see how long people can hold onto the preconceived notions and biases when offered direct evidence against.

@10x_er Hey it's working for ChatGPT

“so, your company is, basically, a good UI with many calls to OpenAI API under the hood.. and that’s what you’re famous for?”

@DeAiDave @thisisbrians @8teAPi Specifically its the number of papers that have received at least h citations - this person has at least 45 papers and those 45 papers have been referenced by others 45 times or more

@thisisbrians @8teAPi H index is a score for reliability. Like Dr. Peter McCullough is the most cited cardiac medical doctor in the world. Anyone can publish, but is your work good enough to be cited? The more it is cited the higher your h index.

They pulled in a physical chemist with 11k citations and a h-index of 45 on the second paper. He signed off on the Meissner effect levitation.

GPT-4 is getting worse over time, not better. Many people have reported noticing a significant degradation in the quality of the model responses, but so far, it was all anecdotal. But now we know. At least one study shows how the June version of GPT-4 is objectively worse than the version released in March on a few tasks. The team evaluated the models using a dataset of 500 problems where the models had to figure out whether a given integer was prime. In March, GPT-4 answered correctly 488 of these questions. In June, it only got 12 correct answers. From 97.6% success rate down to 2.4%! But it gets worse! The team used Chain-of-Thought to help the model reason: "Is 17077 a prime number? Think step by step." Chain-of-Thought is a popular technique that significantly improves answers. Unfortunately, the latest version of GPT-4 did not generate intermediate steps and instead answered incorrectly with a simple "No." Code generation has also gotten worse. The team built a dataset with 50 easy problems from LeetCode and measured how many GPT-4 answers ran without any changes. The March version succeeded in 52% of the problems, but this dropped to a pale 10% using the model from June. Why is this happening? We assume that OpenAI pushes changes continuously, but we don't know how the process works and how they evaluate whether the models are improving or regressing. Rumors suggest they are using several smaller and specialized GPT-4 models that act similarly to a large model but are less expensive to run. When a user asks a question, the system decides which model to send the query to. Cheaper and faster, but could this new approach be the problem behind the degradation in quality? In my opinion, this is a red flag for anyone building applications that rely on GPT-4. Having the behavior of an LLM change over time is not acceptable. Have you noticed any issues when using GPT-4 and ChatGPT lately? Do you think these problems are overblown?

@jesse_hynes Sick

@kaumaron Imagine investing $90,000 in a node and being penalized for trying to recoup more than $20 a month 🤮🤮🤮

I want to remind people of this post in relation to the new Gala Games node vote. Unfortunately the new proposal isn’t any better imo and we as node owners are being forced to take massive steps away from decentralization rather than towards it.

@generativist I've been saying this for like 15 years

hear me out: a constitutional amendment to allow arnold — and only arnold — to run for president

@Will_DeCotiis So much slower

For all y’all who live in a state where you have to pump your own gas… two words for you… New Jersey