My initial impression after reviewing the flight track last night was that the aircraft was experiencing roll control problems and finally rolled left at high speed into the water.
Although the maintenance history of Captain side pitot problems may be relevant, and there is always the possibility of maintenance action causing new problems, an airspeed indication problem of itself should not cause loss of the aircraft.
Those of you speculating on the possibility of stall warnings being triggered should remember that these warnings normally are generated from the angle of attack data which is not derived from airspeed but instead from the vanes on the sides of the cockpit.
The impact of defective air data on the aircraft feel systems needs to be discussed by someone with more 737 aircraft familiarity than myself but could potentially be a significant factor. Aircraft control appeared to be degraded early in the flight.
Speculation: Could both of the cockpit crew have been applying force to the control column and not have had a free hand for setting the power? The continued acceleration of the aircraft doesn't make good sense from a piloting perspective. They needed to set a reasonable power setting and investigate if they could not continue the climb or did not wish to continue the climb.

I'm also disturbed that this document is in the public domain, but having lived and worked in Jakarta, I am not in the least bit surprised.

Pitot cover?
 

The probability of forgetting to remove the pitot covers on a day after there was a write up for Unreliable Airspeed is identical to the probability of forgetting to remove the pitot covers on a day after there was no write-up for UAS.

Not that I'm trying to claim that pitot covers are the cause here, but neither is it a given that the crash is due to an unreliable airspeed indication. That is just one of the few pieces of information that we have, that there was a UAS write-up on the previous flight, and there is speculation how that might have caused the crash, if it were to have reoccurred, whcih isn't a given.

No, the data that is available for download from FlightAware and FR24, while it certainly requires care in interpreting, is what the aircraft was sending (with a couple of caveats that aren't relevant here) and is independent of any prediction algorithms.

One thing that sticks out from the plot is that, aside from the vertical excursions, the last five minutes or so before the plot stops consist of two minutes of dead straight flight followed by two and a half minutes of continuous small track changes. I'll leave it to other to speculate on the significance of that.

I'm currently in Jakarta and can confirm there was a very large storm last night from about 9pm-12am with considerable rainfall

So the prediction you are mentioning is only implemented for displaying live flights, but it is used because quite a few people will look up and want the planes on map to match up.
Also i would not trust the timestamping on the data, the outliers in the following plot are very likely because of one of the receivers clocks being off.

Apart from that FR24 on their blog released the raw data for the flight in much better granularity than available on their website flight history.
It's unlikely that these data, probably received by multiple receivers differ significantly from what was sent by the ADS-B sender of the aircraft in question.

Because the plots of the data on the blog website are somewhat lacking, i plotted the data below in case anyone is interested:

https://cimg3.ibsrv.net/gimg/pprune....f29dd31525.png

ADS-B Data Plot

Thank you for the expanded plot wiedehopf. The final descent appears to be almost vertical with approx 2500' lost in 3 sec.
Condolences to all involved.

OAP

Equal and opposite reactions
 

Fascinating thread, yet deeply tragic.
It seems from the graph that everything has an equal or opposite reaction!
One crumb of comfort would be that the debris field is fairly limited in shallow water and that the CVR and FDR should be recovered promptly and resolved, so that everyone can understand what happened here.
At times like this I can only think of Sir Arthur Conan Doyles' quote
"Once you eliminate the impossible, whatever remains, no matter how improbable, must be the truth."
RIP

Surely the Tech Log would have been on the aircraft?

Presumably. However, it is within the realm of possibility that someone would have photographed the relevant page at some point between blocking in on the previous flight, and blocking out on the accident flight.

Please keep in mind that this is altitude plotted against time so you can't really tell the flight path from it.

If you believe the data the aircraft had a descent rate of 31000 ft per minute during the last recorded transmission.
That is around 300 knots vertical speed. Horizontal speed at that last data point is 360 knots.
That is a total speed of around 470 knots with the flight path angle being around 40 degrees downwards.

So this calculation is kind of gruesome but as the question came up i'll post it anyway.

Salute!

Besides the air data problems on previous flight, and possibly on the accident flight, I would like to re-visit the stabilizer. Just a fast mover guy here, but the sad story is worth understanding. Never know when a pilot here might see something early enough to take action, jez saying.....

Looking at thelfight path plot we have from a previous post, the end seems to happen very quickly, and not a "gradual" loss of control due to PIO chasing the altitude, attitude and speed.

So for you Boeing folks, is there a jackscrew stab trim as we had on the Alaska crash? A squak previously described unusual stab trim. If the stab "came loose" suddenly, I do not see any recovery possible. As Bach said, all crashes are the result of "loss of control". Plane comes apart, big part fails, new FBW laws act up, crew screws up, etc.

Gums asks....

Va= 260 kias at sea level
and 306 kias at 31,000 feet

So, yes you could.

As someone with very limited single engine experience long long ago I'd love to get some thoughts from those in the pointy end. Just looking at the track between 23:25 and 23:29 I'm slightly confused. My understanding is that if the correct memory items where followed you would have established pitch and power as constants...normally in a gentle climb while troubleshooting. During those 4 minutes it appears that both speed and altitude remain "relatively" stable but with clear oscillations that appear to indicate an inverse speed to altitude shift so there is clearly an issue maintaining level flight. I've always thought that at higher altitudes hand flying a swept wing plane was significantly harder than at lower levels. Under an emergency would the aircraft commander normally take the controls and rely on the FO to trouble shoot or would he be more likely to allow the FO to maintain level flight while he looked at the issues?

It would appear that there was an upset and recovery at roughly 23:29:30 and then again roughly 2 minutes later that proved unrecoverable. In both instances it appears to be a roughly 800 ft drop over roughly 20 seconds...that seems awful "ham handed" for purely pitch induced control input to me without some sort of rolling oscillation involved??? What type of flight path variation would be "normal" for an experienced airline pilot with minimal hand flying currency? Hard to imagine losing the plane at 5000 ft. in benign VFR conditions without some other issue in play???