
Old geek with an interest in many things. Hard rule #1 Block me to stop me replying and you get blocked. 100%, no exceptions.
Wherever I am at the time
Joined May 2022
- Tweets32.7K
- Following217
- Followers1.1K
- Likes4.2K
I had to lift my subscription to Ultra. It's costing me more to bridge the weekly allowance than it will cost me for one month of Ultra. I'll put it back when I have done this work.
I'm learning the hard way. "Clone this" means, "sure your file system can handle 1m records easily but the source app had a 32k limit so I limited it at 32k". Trivial to lift but nearly cost me a day of populating my database with the "reading" my local bots are doing.
There is some interesting stuff going on here. My Hermes is burning Openrouter credit on Deepseek V4.1 (the cheapest model on there), where Grok would be emptying everything in a day.
However we made some changes, upped the workers and it cleaned out my openrouter credit in 3 days.
Talked to Bot about it and Bot said, "sure I can modify the python scripts they are running to do much of the decision work only referring back to deepseek if it is not clear". Small work from Grok, reduction from $9 a day to $0.04 a day.
I now have a simple database with a super simple creation and management interface which can handle 1m records, marks every db and record with a GUID and can sync copies with tens of thousands of records in a very short time. Is built in Windows, ported to Android and Linux and the only reason it is not on iOS yet is I don't have the MAC environment to generate it. Issue Fixer will look at that this week.
And, no, I have not written a single line of code and, yes, it all works with thousands of tests already done and confirmed. Sure it burns some of my time testing (although I have a tester bot to reduce my testing to the minimum), however most of my time is spent in making decisions, drawing from other experience I already have in the field and making suggestions on look and feel and how it works.
Next job will be to create new lists (views), which will allow me to view the database with different columns and give far more control on exported or printed output.
The original works had one list, one form, fields tied to columns in the list, no way to scale the print output and no way to hide fields from the list columns.
We are now into about 1.5 man years of an average dev. Bot can do stuff faster than I can even open a reference page to see how to do it.
I guess the final challenge will be making something far more powerful than this but that will take a whole new level of my time.
46
NeilT retweeted
Replying to @PebMet1
If the shuttle were half of what is designed for Starship I might agree. The shuttle was supposed to be "quickly" reusable, not "rapidly". It never came close.
Attempts to speed it up led to the booster failure I woke up to one morning. Trying to speed up the central tank, fully disposable, also led to the loss of crew.
The Shuttle was rushed out, even in the 9 years, they veto'd an escape capability due to cost and then lost crew through "rushing it".
We saw similar just recently when they tried to force a return of crew with Starliner after having nearly lost the crew on ascent and ISS docking and doing Boeing some rapid patching of a highly irregular nature.
Starship aims for rapid reusability, safe crew transport, 200t to 400t to orbit depening on reuse and it will supply crew and cargo transit to both the Moon and Mars.
The shuttle was nothing more than a LEO transit which was supposed to build space stations and transfer crew to them at 70 at a time or more.
During the ISS build it proved too slow to launch to meet the ambitions of multiple large space stations and too dangerous to trust with large flights of civilians.
You then say "9 years to first crew" and assume it is a comparison.
In fact a rocket of this type would have taken NASA decades to produce. If ever.
If people want to make comparisons they need to create a list of goals for the shuttle design and then the achievements against those goals. Then compare that to Starship goals and where it is.
It is also worth remembering that the shuttle could not fly itself. So the first flight had to be crewed or it would not happen. The same high risk situation it continued in throughout its career.
Congrats SpaceX and next time it will be both.
Incredible achievement and the haters are wallowing in their own bile.
54
NeilT retweeted
Replying to @Eddystone506
So it had nothing to do with massive uncontrolled spending during Covid and war in Ukraine and Iran which sucked the rest of the world in and dramatically reduced Oil supply.
It is amazing to know that the UK So powerful that it caused 5% global inflation just by leaving the EU.
Good to know you are clued in as to what cases inflation.
17:39 to my biluilder bot.
The clone of works DB we have made, I would like a full port to Linux.
18:02, Builder bot.
"WorksDB for Linux 2.9.0 is built and tested, and it has almost the full Windows feature set: List, Form and Report views, Form Design with multi-page forms and =FieldName labels, the List dropdown, search, sort, filter and mark, attachments, Print to PDF, the List heading font, and full sync. There are x64 and ARM64 builds. Files work interchangeably with Windows 2.8.3 and Android 0.5.0. "
Now time for the tester bot then whatever is left for me to test.
97
Replying to @matthewstoller
Left: Elon is successful only because of government support.
Elon: Fine, I won't take any government money and cancel govt contracts.
Left: How dare you threaten me.
I ran out of Grok Bot tokens again. So I decided to try a stop gap of buying some tokens to continue.
Grok Bot uses tokens like a mad Cossack drinks Vodka.
However the local LLM is up, the local agents are rectified and connected, now All I need to do is get Grok to orchestrate which should reduce tokens a lot.
1
5
134
Side take.
When the Dems realize how much of the future economic growth in the US will depend on AI, they will U turn and support it.
1
3
77
Possible take.
SpaceX has recently dramatically improved the android experience for X and for Starlink.
Why we have to wonder, it has been Apple first last and always, with Android getting tough love.
SpaceX intend to provide direct to mobile with their own frequencies. Frequencies that are curre tly not supported by handsets. Which means hardware development to support it.
What has Apple sucked at for a long time now?
Truly innovative and ground breaking hardware.
Not so much for the Android handsets.
Just a thought.
NeilT retweeted
Replying to @Fernandonyny
The only reason Ukraine still exists to attack Russia is because Elon workers with the Canadian and German governments to get Starlink to Ukraine when it was communicate or lose.
Biden and his sycophants were dragging their feet and not helping. They would rather have seen Ukraine lose to Russia than Elon get one good word in praise.
They you sit there like a lying sack of shite and troll out your lies and call anyone who diflsagrees with you a Russian bot.
You are the lying trolling left bot and are Butthurt because Elon exposed your financial graft and left the US government no choice but to cut it off.
Don't bother blocking me, I reposted this.
Still winning with Grok Bot.
No sooner have I managed to get a cleansheet of Works Database, but upgraded, rich text, objects, hidden fields on the view, but still with most of the power features, I say "right can I have an android version, no design just view and data entry".
2.5 hours, first android version, see it all, no filters yet, no reports. But simple mostly view only with a bit of data entry.
Whilst I was doing all of this I had another bot fill in all the isbn's and blurbs and cover art for 2,500 books in the library db and I had it scour 1,500 recipies I had in pdf's and other documents and build it out with description, pictures if any, up to 25 ingredients with the rest (too many ingredients sometimes), below the rich text instructions.
All available also on the tablet. Bring it to the kitchen stick it on the stand and start cooking.
I will include selecting and only viewing certain documents to make that easier when cooking several recipe's but 2.5 hours.... If I were to look at Android programming it would be 2.5 years to pick it up.
Next a port to iPad as my wife has an iPad, I have androids.
Final key feature will be back end server sync. but that is probably after Christmas.
Bot doesn't always get it right but I have a tester bot and then I do the final few tester is not happy with.
Bot is more than paying for itself. I have a full time job doing other things and haven't core programmed in over 25 years.
2
79
Still winning with Grok Bot.
No sooner have I managed to get a cleansheet of Works Database, but upgraded, rich text, objects, hidden fields on the view, but still with most of the power features, I say "right can I have an android version, no design just view and data entry".
2.5 hours, first android version, see it all, no filters yet, no reports. But simple mostly view only with a bit of data entry.
Whilst I was doing all of this I had another bot fill in all the isbn's and blurbs and cover art for 2,500 books in the library db and I had it scour 1,500 recipies I had in pdf's and other documents and build it out with description, pictures if any, up to 25 ingredients with the rest (too many ingredients sometimes), below the rich text instructions.
All available also on the tablet. Bring it to the kitchen stick it on the stand and start cooking.
I will include selecting and only viewing certain documents to make that easier when cooking several recipe's but 2.5 hours.... If I were to look at Android programming it would be 2.5 years to pick it up.
Next a port to iPad as my wife has an iPad, I have androids.
Final key feature will be back end server sync. but that is probably after Christmas.
Bot doesn't always get it right but I have a tester bot and then I do the final few tester is not happy with.
59
NeilT retweeted
Replying to @SciGuySpace
I asked Grok to summarise what I read in the article. Grok leans towards accuracy and fairness so it is not my "hit piece".
The final summary.
The piece is strongest when it documents the concrete contract pullbacks and the resulting NASA dependency. It is weakest when it treats current low Starship flight rates and higher NASA refueling estimates as near-permanent features, downplays production and architectural commonality with HLS, and applies a more generous “they are moving fast and will get there” filter to Blue Origin than to SpaceX. Your analysis tracks the article’s actual claims and the places it chooses not to look.
The rest.
Yes — the piece is structured as a long, carefully framed cautionary tale that treats SpaceX’s rational pivot away from Falcon/Dragon as a national crisis, while giving Blue Origin’s promises more weight than SpaceX’s existing hardware trajectory. It is not a pure hit piece, but the selection of facts, omissions, and tone line up closely with your reading.
Core framing and tone
Berger opens with the “El Dorado” metaphor (SpaceX hacking through the jungle of Boeing/Lockheed to win government contracts) and closes with the need for “another billionaire” (Bezos) to bail out NASA. The article’s title and subhead (“NASA desperately needs SpaceX, and SpaceX knows it, and Elon knows it”) set a dependency narrative. The Trump–Musk contract-threat episode is presented as proof the power dynamic has inverted, with commentary that Musk’s Dragon threat was “utterly irresponsible.” The political pivot you noted is explicit in the conclusion: NASA “turned its spaceflight fortunes over to one billionaire” and may now need another.
The piece repeatedly notes that SpaceX is still fulfilling existing contracts and remains more reliable/cheaper than alternatives, then pivots to the risks of the company choosing not to sign new traditional launch work. That is accurate as far as it goes (no new commercial Falcon sales after 2028, Crew Dragon wind-down by ~2030, declining some Space Force and CLPS-related launches), but the surrounding language treats the choice as abandonment rather than a company whose revenue is now dominated by Starlink + AI/compute and which is concentrating engineering on Starship.
Starship production, cadence, and HLS relevance (the gaps you flagged)
- Berger correctly states Starship has flown 14 times in ~3.5 years and that even a weekly cadence would make a ~20-launch refueling campaign take months. He notes SpaceX’s longer-term FAA filing for very high rates (10,000/year by the 2030s) and Musk’s 6–8 figure versus NASA’s higher near-term planning number (~22 / “about 20”). He also notes that higher performance and better refueling operations will reduce the number.
What is largely absent, as you said:
- Current production and test activity (multiple vehicles in test, rapid stacking, Megabay/Gigabay infrastructure at Starbase plus Florida capacity coming online).
- That HLS is a conversion kit on the same core architecture (rings, tanks, Raptor count, structural approach) rather than an unrelated new vehicle. Flight-rate progress, engine reliability, and structural/thermal work directly feed the lander.
- Pad and recovery differences versus Falcon. Falcon’s three launch mounts, drone-ship recovery cycle, and high rate of expended Falcon Heavy center cores are real constraints. Multiple Starship pads with higher reuse potential change the annual throughput math quickly once the vehicle is flying regularly. Your point that five pads with moderate reuse could exceed recent Falcon annual records (approaching or exceeding ~160) is a reasonable near-term scaling argument that the article does not engage.
On payload density for tankers versus Starlink: the article leans on current partial Starlink loads and the higher NASA planning number. It does not address that propellant is far denser than a stack of satellites in the same ~1,000 m³ class volume, that dedicated tanker ships with internal tanks are the logical architecture, or that single-use options (even a few of them) would further cut the flight count. Those are fair engineering observations the piece does not engage.
Blue Origin asymmetry
The article notes New Glenn’s pad explosion, then emphasizes the “uncharacteristic speed” of the rebuild, the $3 billion Florida investment, and the possibility that New Glenn becomes “the essential rocket powering NASA’s lunar ambitions.” It correctly notes Blue Origin’s short orbital track record (three launches) and that the lander does not yet exist. It still frames Blue as the plausible near-term alternative and Bezos as the billionaire who would “relish the role.” That is the optimism-versus-hardware contrast you highlighted: real flying hardware and production infrastructure at SpaceX versus a vehicle that recently failed on the pad, a lander that is still conceptual for Artemis III timelines, and a company whose execution record is shorter.
Additional alignments
- The Memphis / terrestrial AI reallocation is presented as evidence of deprioritization of Starship. It is also consistent with a company that has multiple large engineering problems (Starship flight rate + orbital compute) and is allocating talent accordingly.
- The financial comparison (high near-term Starship launch costs times 20 flights versus the original HLS contract value) is used to question incentives. It is less persuasive once higher reuse, higher payload density, and the fact that Starship is primarily an internal-payload vehicle are factored in.
- Several public replies to the X post (Handmer, Robotbeat, others) make versions of the same points: SpaceX can now say no; flight rate is the shared bottleneck for Starlink/AI and HLS; the industry has a long history of underestimating SpaceX cadence once the vehicle works.
Grok Bot keeps excelling.
I get a reminder monthly for a free service I use. I have to log in, select a link and click on it, then in the next screen click to confirm to continue.
It's a pain but it is meant to try and make me subscribe to make the pain go away.
I just told Bot where to go, we logged in and vaulted the credentials, I navigated the MFA and Grok simply said (It's done).
Duly dubious I logged in and checked because Grok had, as far as I know, not seen the page before.
And it was confirmed.
Now Grok Bot will ask me, monthly, at around renewal time, to present my MFA and it will log in and check and confirm if it needs to be done.
The days of me forgetting and having to set things up again appear to be gone.
1
57
And in a really good F.U.2 to Ryanair, Easyjet take Starlink.
Over time this is going to burn. Especially as Easyjet incorporate this into their advertising.
For those who don't know Easyjet is a major competitor to Ryanair in the low fares segment.
Starlink will deliver fast, reliable connectivity onboard @easyJet, so passengers can work, stream, or stay in touch without interruption 🛰️✈️
NeilT retweeted
Replying to @CristianaEmsley
The rebate is gone, if we go back it isn€20bn a year. But the new budget isndiloubke the old one and they never shrank it after we left.
So around €40bn a year.
If we have that much money to waste, why are we not spending it on the NHS.
They say we don't have nough money but they want to pour this money down and endless pit.
We have the true figures for net contributions now, they came out in the last EU budget negotiations when they had to fill the hole.
So we should be asking them why, if they have this money to throw away, why they are not spending more on the NHS and why are they shouting to tax more?
NeilT retweeted
Replying to @LizWebsterSBF
Excellent, just waiting for that EU parliament vote to vote in the biggest asshole who can do the most damage to the EU with all those others who are gradually starting to win majority in the parliament.
So we can be an even bigger pain in their ass then last time.
Do enjoy paying into their doubled budget which they will decide before we rejoin and we will have no chikce but to pay, €27bn a year. That's a damned sight more than 340m.
Just remember. We were the b'stards of the EU and a great many were glad to see the back of us.
Oh and they got rid of the rebates after we left. €40bn a year coming soon to an EU near you.
Do enjoy because we're going to love telling them all how wrong they are.
All over again.
NeilT retweeted
Replying to @Suewilson91
No you want to drag us back into that mess for your own self interest.
So get this and understand.
There are more of us than you
We don't like the EU, most little Englanders don't like furriners either.
We will vote Reform in the EU parliament elections to deliberately screw up the parliament and if they won't do it we will vote for Restore who certainly will. Because the EU parliament is a "safe" vote, it doesn't matter.
We will vote for MP's who say they will "stick it" to the EU and we will vote for parties which can break the balance in Westminster and destabilise the country. Because if we break it we can win on marginal politics.
We Will come to Europe and tell them they are all wrong. Our MEP's will stand up in their parliament and tell them we don't like them and if they don't change we will work with the other, growing, disaffected groups and create havoc.
We will be the largest pain in their butt they have ever seen and that includes the last time so that they think the money isn't worth it.
Oh you didn't know this is all about that blockbuster new EU budget they can't afford?
What we are going to do is be exactly what we were before.
Just more.
Because now we know we don't have to put up with their crap and we can get out.
We are going to be so bad that they will write a new article, one which says they can force a member out of the Union. Then we are going to sign it into law as fast as we can.
Then we will have two ways out.
You may get your selfish deserts and Burnham may get his phyrric victory. But the EU will Rue the day.



